External Costs of Truck and Rail Freight Transportation

Transportation & Vehicle Safety Policy

1-1-1998

External Costs of Truck and Rail FreightTransportationDavid J. ForkenbrockUniversity of Iowa

Copyright © 1998 by the Public Policy Center, The University of Iowa.

Hosted by Iowa Research Online. For more information please contact: [email protected].

http://ir.uiowa.edu?utm_source=ir.uiowa.edu%2Fppc_transportation%2F2&utm_medium=PDF&utm_campaign=PDFCoverPages



https://ir.uiowa.edu/ppc_transportation?utm_source=ir.uiowa.edu%2Fppc_transportation%2F2&utm_medium=PDF&utm_campaign=PDFCoverPages

http://ir.uiowa.edu

mailto:[email protected]

External Costs of Truck and RailFreight Transportation

External Costs of Truck and RailFreight Transportation

David J. ForkenbrockDirector, Public Policy Center

Public Policy CenterThe University of Iowa1998

This study was funded by the University Transportation Centers Program of the U.S.Department of Transportation and the Iowa Department of Transportation. Theconclusions are the independent products of university research and do notnecessarily reflect the views of the funding agencies.

ii

Copyright © 1998 by the

Public Policy CenterThe University of IowaIowa City, IA 52242

All rights reserved

iii

PREFACE

Comparatively low-cost freight transportation has been an important element in thegrowth of the U.S. economy. Goods can be transported between most points in thecountry quite cheaply and efficiently. To varying degrees, however, the freighttransportation services we consume generate costs that are borne by others. Suchcosts are commonly referred to as external costs.

From a societal perspective, it is desirable for all transportation services to pay theirfull social (private and external) costs. If the full social cost were reflected in theprices shippers pay, transportation users could choose the amount of each form ofservice to consume on the basis of the true cost of this service to society. By“internalizing” external costs, policy makers would effectively create a marketthrough which transportation users could weigh the benefits of consuming aparticular transportation service against the true costs. The purpose of this monographis to estimate these true costs for freight truck and rail.

We estimate four general types of external costs for a ton-mile of freight shipped bytruck or rail: accidents (fatalities, injuries, and property damage); emissions (airpollution and greenhouse gases); noise; and unrecovered costs associated with theprovision, operation, and maintenance of public facilities (primarily roads andbridges). Because the preponderance of freight transportation occurs between cities,we focus on intercity freight flows and ignore the movement of goods within urbanareas. Consequently congestion, a primarily urban phenomenon, is not addressed. Anintercity focus also simplifies the estimation of air pollution costs. Whereas pollutionlevels in rural areas are fairly consistent, an additional unit of pollution can bringabout costs that vary greatly among metropolitan areas based on existing air quality.Our analysis thereby serves as a benchmark against which more specific externalcost estimates can be compared.

The research reported in this monograph was carried out at the University of IowaPublic Policy Center. Funding was provided by the U.S. Department ofTransportation, University Transportation Centers Program, with supplementalfunding contributed by the Iowa Department of Transportation.

v

ACKNOWLEDGMENTS

In the preface I mention that this research was funded by the UniversityTransportation Centers Program of the U.S. Department of Transportation and by theIowa Department of Transportation. Both of these agencies have my gratitude fortheir support.

Data needs for this project were extensive, and completion of this study would havebeen much more difficult without the assistance of several people. In particular, LisaSchweitzer, my colleague at the Public Policy Center, was highly enterprising andthorough in helping me find necessary data. Harry Cohen provided data on airpollution costs in rural counties, and Scott Dennis shared various data from theAssociation of American Railroads. Alan Krupnick advised me on informationsources related to the rates and costs of air pollution.

I acknowledge the important contribution of Gregory Bereskin, who modeled fourfreight railroad cost scenarios. It was a complex task to estimate the private costsper ton-mile of the four different railroad configurations. His work has enabled me tocompare external costs with private costs for representative types of freight railoperations. A draft of this monograph was reviewed by the following people:Gregory Bereskin, Scott Dennis, Sheffer Lang, Sondip Mathur, Robert Pitcher,Thomas Pogue, and Lisa Schweitzer. They provided numerous suggestions thatimproved the monograph in many ways. Any errors that remain, of course, are myresponsibility.

Anita Makuluni, editor at the Public Policy Center, ensured that the text isaccessible to a wide audience, while maintaining the monograph’s technicalaccuracy.

With real appreciation I acknowledge the many and diverse contributions of thesepeople.

vii

TABLE OF CONTENTS

PREFACE.........................................................................................................iii

ACKNOWLEDGMENTS...................................................................................v

FIGURES ....................................................................................................... ix

TABLES..........................................................................................................xi

CHAPTER 1: INTRODUCTION.........................................................................1

External costs and benefits .......................................................................1

An intercity focus....................................................................................2

Modal competition..................................................................................2

CHAPTER 2: PRIVATE COSTS...........................................................................7

Motor carrier operating costs ....................................................................7

Freight rail operating costs .......................................................................9

Summary.............................................................................................. 12

CHAPTER 3: NON-MARKET COSTS................................................................ 13

Accidents............................................................................................. 13

Emissions............................................................................................. 20

Noise .................................................................................................. 28

Conclusions.......................................................................................... 31

CHAPTER 4: GOVERNMENT SUBSIDIES......................................................... 33

Costs occasioned by vehicles ................................................................. 33

Do freight trucks underpay?.................................................................... 35

Conclusions.......................................................................................... 40

CHAPTER 5: INTERNALIZING EXTERNAL COSTS............................................ 41

Private and external costs....................................................................... 41

Conclusions.......................................................................................... 45

REFERENCES................................................................................................. 47

ix

FIGURES

1–1. Competitive freight service for truck and rail, 1994.......................................3

5–1. ITS approach to charging heavy vehicles for road use .................................. 44

xi

TABLES

2–1. Private operating costs of truckload (TL) general freight trucking, 1994...........9

2–2. Private operating costs of four railroad freight scenarios, 1994..................... 12

3–1. Cost of accidents .................................................................................. 15

3–2. Accident rates and costs of passenger cars and large trucks, 1994 ............... 16

3–3. Compensation for accident costs paid by truckload general freighttrucking companies, 1994....................................................................... 17

3–4. Costs of accidents involving Class I freight rail, 1994................................ 19

3–5. Compensation for accident costs paid by Class I freight railroads, 1994 ....... 20

3–6. Types of air pollution............................................................................. 22

3–7. Average air pollutant costs for 2,233 rural counties.................................... 23

3–8. Emission rates for selected truck configurations......................................... 24

3–9. Emission rates and costs for freight trucks................................................. 25

3–10. Emission rates for four types of freight trains............................................. 26

3–11. Emission costs of four types of freight trains.............................................. 26

3–12. Rates and external costs of carbon dioxide emissions, truck and rail, 1994 ... 27

3–13. Decibel levels based on traffic volume and distance from the road.............. 29

3–14. Effects of noise levels............................................................................ 29

3–15. Summary of external costs of truck and rail freight .................................... 32

4–1. Estimated distribution of federal highway obligations in year 2000 .............. 34

4–2. Highway obligations or expenditures for 1994 and year 2000 at alllevels of government ............................................................................. 36

4–3. Year 2000 highway user charge payments by vehicle class at all levelsof government....................................................................................... 37

4–4. Year 2000 federal cost responsibility for all levels of governmentby vehicle class.................................................................................... 38

4–5. Ratios of year 2000 user charge payments to allocated costs for all levelsof government....................................................................................... 39

4–6. Underpayment of highway user charges by vehicle class for all levelsof government, year 2000....................................................................... 40

5–1. Private and external costs of truck and rail freight ..................................... 42

CHAPTER 1INTRODUCTION

Freight transportation is a vital element in the economies of nations, regions, andcities. Low-cost, dependable movement of freight helps a business to be competitive.Location of facilities, choice of shipment size and mode, and competitive biddingare among the ways a business seeks to keep its (private) transportation costs as lowas possible. In the same vein, it is good public policy for society to try to minimizeits total transportation cost, while ensuring that people and goods are movedeffectively. First, however, policy makers must identify the full social costs fordifferent modes of transportation so policies can be adopted that encouragetransportation users to consider these costs when making travel or shipping decisions.

Ideally each unit of transportation service used (e.g., a person-trip or a ton-mile offreight) could be assigned a price that would reflect the full incremental cost tosociety of that unit of consumption. Charging the full incremental or marginal costsof transportation would establish a market in which transportation users could decidewhether the benefits to them of consuming a particular unit of transportation wouldexceed the costs these users face. In economics parlance, such a policy would leadto efficient use of transportation services. It also could lead to equitable pricing oftransportation, such that society would not absorb the costs of one mode or type ofservice to a greater extent than another.

EXTERNAL COSTS AND BENEFITS

External costs (or negative externalities) must be added to private costs to arrive atfull social costs. Thus, to achieve full social cost pricing of transportation services,one must first comprehensively and accurately estimate relevant external costs.Varian (1984) defines external costs thus:

In the basic general equilibrium model economic agents interactonly through their effect on prices. When the actions of one agentaffect the environment of another agent other than by affectingprices, we say that there is an externality (p. 259).

By this definition, external costs and benefits are outside normal market processes(i.e., are not reflected in prices). Externalities constitute a form of market failurebecause true costs are not taken into account when production and consumptiondecisions are made. If external costs are greater than external benefits, notconsidering externalities may lead to over-consumption of transportation.

Significant external costs arise from transportation, although accurate cost estimateshave been difficult to obtain. In this monograph we assemble best available

2 EXTERNAL COSTS OF TRUCK AND RAIL FREIGHT TRANSPORTATION

estimates of external costs due to freight truck and rail operations between cities(i.e., in rural areas). To place these costs in a useful context, we compare them withcorresponding private costs, which are the market costs directly faced by carriers.

This monograph does not address the issue of external benefits that arise through thetransportation of freight. Most transportation economists doubt the existence ofsignificant external benefits from the provision of transportation services (e.g.,Rothengatter 1994; Verhoef 1994, p. 278). Markets exist for these services, such thatcost savings due to economies of scale or density may be retained by carriers, passedon to shippers, or ultimately passed on to final consumers of the shipped goods.

As Greene and Jones (1997, p. 9) observe, some economists believe that externalbenefits can arise from improvements to transportation systems (see, for example,Willeke 1994). An improvement may reduce the costs of firms’ operations, thuscontributing to increased competitiveness and more output. If scale economies exist,the unit cost of production may drop. Depending on market dynamics, thebeneficiaries of this cost reduction may be the producing firm or its customers. In thisanalysis we focus on transportation services rather than facility investment andtherefore do not consider external benefits.

AN INTERCITY FOCUS

Our objective in this monograph is to bring together what is known about theexternal costs of shipping freight between cities via truck and rail. We focus onintercity freight movements for two primary reasons:

• By far the largest number of vehicle-miles and ton-miles of freighttransportation occur between cities, as opposed to within cities.

• External cost levels generally vary much less among rural areas thanamong cities. For example, ambient air pollution levels may beappreciably higher in one metropolitan area than another. In a citywhere pollution levels are already relatively high, additional truck- orrail-generated pollution will add greater social costs. Likewise,congestion costs are negligible in rural areas but range from low toquite high in U.S. cities. Applying an aggregate estimate ofcongestion costs when these costs are so variable would degrade theusefulness of the analysis.

Our general framework is amenable to adjustment for external costs prevailingwithin a particular metropolitan area or region. Good estimates of specific accidentrates, congestion costs, and appropriate cost values for air pollutants would make itpossible to estimate social costs for such an area.

MODAL COMPETITION

Under full social cost pricing of freight transportation modes, the true costs to societywould be reflected in the prices paid by users, and therefore the modes would be

Introduction 3

able to compete on an equal basis. How the inclusion of external costs would affectmodal competition between rail and trucking would depend on a number of factors,including relative service quality and the extent to which the two modes are able toserve the same markets. In general, rail and trucking compete in markets involvingdistances that are relatively short for rail yet relatively long for trucking. Most often,the value (dollars per ton) of freight shipped by truck is higher than that shipped byrail. One must recognize that our general unit of analysis, the ton-mile, includes avery wide array of goods.

Figure 1–1 indicates the amount of freight (measured in ton-miles) shipped in theUnited States by long-haul truck and freight rail in 1994. Of particular interest arethe shaded portions of both pie charts: 41 percent of long-haul truck ton-miles arecompetitive with rail, and 33 percent of rail ton-miles are competitive with truck(Abacus Technology Corp. 1991, Exhibit 5–1). In total, about 768.5 million ton-milesshipped annually are modally competitive.

41%

59%

67%

Competitivewith rail

Not competitive with rail

Total: 1,200.7billion ton-miles

Total: 908.0 billion ton-miles

Long-haul truck

Freight rail

Ton-miles (millions)

Commonand competive commodities

Competitive with truck

Not competitivewith truck

Total:768.5

396.2(56.9%)

Truck

Rail

33%

372.3(43.1%)

Figure 1–1. Competitive freight service for truck and rail, 1994

SOURCES : Bureau of Transportation Statistics (1996a, pp. 41, 53). Percentage modalcompetitive estimates are for 1987 from Abacus Technology Corp. (1991, Exhibit 5–1).Freight rail ton-miles are for Class I railroads.


If full social cost pricing were to become policy, the extent of any resulting shift inmodally competitive freight in a given market would depend on several factors,including:

• the magnitude of change in relative prices for various types ofshippers,

• the difference in quality of service provided by competing modes, and

• specific requirements on the part of shippers.

Aggregate estimates of these factors would be difficult to make. Thus, the change inmodal shares if full social cost pricing were in effect can only be speculated on,even if the magnitudes of price changes were known. Our interest in this monographis to estimate the size of external costs and the extent to which full social costpricing would exceed current market costs faced by freight rail and trucking carriers.It is not our objective to argue for greater use of one mode or another.

Marginal versus average costs

In estimating social costs, ideally we would examine the marginal cost to society ofone more unit of freight transportation service. If a freight carrier pays marginal usercharges that equal the marginal social cost of the unit of freight, the provider oftransportation service is paying appropriately, from a societal perspective.

As the Transportation Research Board (TRB 1996, p. 2) observes, a marginal costperspective is quite different from that used in highway (and other) cost allocationstudies. Such studies are intended to determine how the costs of providinggovernment facilities and services should be distributed equitably among differentvehicle classes. In contrast, a marginal cost perspective is concerned only withwhether the costs borne by society are fully assigned to those generating them.Marginal social cost pricing may be equal to, higher than, or less than the budgetarycost of government for providing facilities and services.

As a practical matter, it is difficult to develop accurate estimates of the marginalsocial costs of freight transportation. For example, good data are available on thenumber of fatalities and personal injuries associated with 100 million miles of truckoperations. Thus, the average accident cost per vehicle-mile can be derived, andusing average load factors, average cost per ton-mile can also be calculated. Themarginal accident cost of one more truck vehicle-mile or ton-mile is much moredifficult to estimate. Trip-specific considerations such as traffic volume on theroadway, design of the roadway itself, weather conditions, and factors peculiar tothe truck and driver all enter the picture.

Estimates of marginal social costs are most valid when they pertain to very specificcircumstances. In its report on marginal social costs of freight transportation, TRB(1996) used four specific case studies and stressed the limitations of these studies inmaking general conclusions about marginal social costs. TRB recommended (p. 125)

Introduction 5

an expanded array of case studies to increase what is known about the social costsof freight transportation.

In this analysis we use average costs largely derived from aggregate data. While ourestimates lack the precision of a more specific case study, these estimates providean overall sense of the magnitude of various types of external costs generated byfreight trucks and rail relative to average private costs nationally.

Application of this research

Our objective is to provide an overall assessment of external costs arising from truckand rail freight transportation. If public policies were formulated to internalize theseexternal costs in an aggregate sense, some carriers of either mode would overpay,while others would underpay. The amount of overpayment or underpayment woulddepend on the difference between average costs in the aggregate and the marginalcosts in a particular circumstance.

In short, unless one is able to accurately estimate the marginal social costs of eachunit of transportation (e.g., each ton-mile) in widely varying circumstances, twochoices are possible. One is to ignore external costs and estimate user charges andtaxes solely on the basis of public facility use, and the other is to accept a degree ofcross-subsidization within each transportation mode. We tend toward the secondoption: developing conservative estimates of average social costs in rural areaswhere the variation in these costs is less likely to be sizable than is the case inmetropolitan areas. Our approach thus can serve as a benchmark against which case-specific estimates can be compared.

Overview

To provide a context for our estimates of social costs, in Chapter 2 we developestimates of the private operating costs faced by freight truck and rail carriers.Although the only important issue from a public policy perspective is the magnitudeof external costs (which need to be internalized), private cost estimates enable oneto gauge how sizable external costs are relative to private costs.

In Chapter 3 we estimate average non-market or external costs for a ton-mile offreight shipped by truck or rail. The three major categories of external costs weinclude are accidents, emissions (both air pollution and greenhouse gases), and noiseimpacts.1 We begin by defining our methodology for estimating accident externalcosts, which are equal to total costs to society of accidents per unit of service minusany compensation paid by the freight carrier. Then we provide estimates of unit costsof fatal, personal injury, and property-damage-only accidents.

1 We use the term “accidents” even though in much of the current safety literature the term

“crashes” is more common. On-the-job injuries to employees or pedestrians being struck bya train or truck are better described as accidents than crashes.


Our analysis of air pollution costs is based on a review of previous studies that haveassigned dollar values to units of various pollutants. Values we obtained are oftenquite wide-ranging, reflecting both differences in conditions at the sites being studiedand uncertainties regarding the appropriate values to assign. We estimate pollutionrates for general freight trucks and the four train scenarios, as well as the resultingcosts per ton-mile. A comparable analysis is presented for carbon dioxide (CO2), themost important greenhouse gas generated by diesel engines.

Finally, in Chapter 3 we explore external costs associated with the noise emitted bytrains and trucks operating in rural areas. Following a discussion of noise effects, wereview previous studies that assessed the economic costs of noise. Lacking betterdata on residential densities, a key factor in noise cost estimation, we use a ratherlow estimate developed in the 1980s. We then apply conservative values of noisecosts for trucks and freight rail.

Chapter 4 contains an analysis of the extent to which freight trucks pay the full costsassociated with their use of public facilities. Because essentially no freight railoperations use public facilities, the chapter addresses trucking only. Using results ofthe 1997 Federal Highway Cost Allocation Study (FHWA 1997a), we estimate thedifference between user charges paid by heavy trucks and the costs these vehiclesoccasion. Our estimates include payments and costs for federal, state, and locallevels of government.

In Chapter 5, we synthesize the results of the previous chapters and drawimplications for trucking and rail.

Private Costs 7

CHAPTER 2PRIVATE COSTS

To determine the fractional increase in freight costs that would result if freighttrucking and rail were to pay the full costs they occasion, it is necessary to estimateboth the private and external costs of the two transportation modes. Private costs arethe direct expenses incurred by providers of freight transportation. Such costs consistof operating costs, as well as investments in capital facilities and rolling stockwhich eventually wear out and must be replaced. Operating costs are those that areclosely linked to the amount of service provided: fuel, wages, maintenance, usercharges, depreciation, and insurance.

External costs are the result of day-to-day operations, so operating costs are the mostappropriate basis for comparisons with external costs. Taken together, privateoperating costs and external costs can give both shippers and carriers signalsregarding the true (full) cost of a unit of service. In turn, the amount of servicedemanded at this cost will define the appropriate level of capital investment.

Because most data on production costs for freight transportation are averages, thesedata will differ from the cost at the margin by the magnitude of any remaining long-run economies of scale. In the discussion to follow, we consider the extent to whicheconomies of scale exist in freight trucking and rail. We then estimate average costsper ton-mile for the most common type of intercity trucking operation and fordifferent types of freight train configurations.

MOTOR CARRIER OPERATING COSTS

Types of motor carriers

There are two basic forms of freight truck service operating in the United States:truckload (TL) and less than truckload (LTL). TL services generally transport ashipment of freight from a single shipper to one or more receivers; freight is pickedup in a line-haul combination truck at the shipper’s dock and transported to thedestination in the same vehicle. TL carriers rarely handle freight at their ownfacilities.

LTL trucking serves many shippers that often send small shipments to be delivered tomultiple receivers. To serve numerous shippers, LTL carriers maintain strategicallylocated terminals. Smaller trucks bring freight to the terminals and distribute it fromterminals; line-haul combination trucks move the freight between terminals. Becauseof the large investment in terminal operations, entry into LTL operations is difficult.


As a result, several large companies dominate LTL trucking (e.g., United ParcelService, Yellow Freight, Roadway Express, and Consolidated Freightways).

The TL market is much easier to enter because all that is needed is a driver, rollingstock, and a freight broker with whom to work. Accordingly, the TL sector is highlyfragmented, being composed of many small and medium-sized carriers. Followingderegulation of the trucking industry in 1978, the number of TL carriers grew rapidly,up from about 17,000 certified carriers in 1979 to roughly 38,000 in 1987 (TRB 1989,p. 73). During this time the number of Class I carriers (annual revenues of more than$10 million) actually fell, while smaller Class III carriers (annual revenues less than$3 million) more than doubled. According to TRB (1989, pp. 70–71), sincederegulation TL carriers have accounted for a steadily increasing portion of the ton-miles of service provided by the trucking sector. By the late 1980s, TL carriers weretransporting over 90 percent of the ton-miles shipped by truck. We focus on TLtrucking in this analysis.

Private operating costs of TL carriers

TL carriers can be divided into six categories:

• general freight,

• automobile transport,

• refrigerated,

• bulk commodity,

• tank truck, and

• other specialized.

For each of these categories, the American Trucking Associations (ATA) compilesoperating and financial data. ATA data include nearly all Class I carriers, some ofthe Class II carriers, and almost none of the Class III carriers. Available private costdata on TL operations therefore tend to pertain to larger trucking firms. The lack ofdata on smaller carriers could bias our cost data downward, to the extent that scaleeconomies exist in the trucking industry. Button (1993, pp. 74–75), however, doubtsthat there are significant increases in returns to scale, citing evidence that largecompanies compete directly with one- or two-vehicle firms. He further notes thatwhile differential managerial skills may permit some firms to grow larger thanothers, that does not in itself reflect scale economies of a technical nature.2 To theextent that there are no sizable scale economies in TL operations, average operatingcost data provide a reasonable approximation of the marginal private cost of onemore ton-mile of service.

2 McMullen and Stanley (1988) conclude that prior to deregulation of the trucking industry in

1980, the industry did have increasing returns to scale; after deregulation the industry hasexhibited essentially constant returns to scale.

Private Costs 9

By far the largest TL category in terms of ton-miles transported is general freight.Because of its dominance and our desire to avoid unnecessary aggregation, we limitour analysis to general freight trucking. ATA breaks down TL general freight truckingby the length of haul carried out by various motor carrier firms. In Table 2–1 wepresent 13 expense categories and key performance measures for three lengths ofhaul. Overall, in 1994 TL general freight trucking had a per-mile operating cost of$1.25, a cost per ton-mile of 8.42 cents, and an average load of 14.80 tons.

FREIGHT RAIL OPERATING COSTS

Estimating private costs of freight rail service is inherently more complex than is thecase with trucking. Among the complicating factors are joint production among railcompanies (e.g., sharing trackage or rolling stock), economies of scale and density,and a lack of data on specific expenditures pertaining to individual freight

Table 2–1. Private operating costs of truckload (TL)general freight trucking, 1994 (thousands of 1994 dollars)

Length of haul

Expense categoryUnder

250 miles250 to

500 milesOver

500 milesAll TL generalfreight carriers

Salaries 46,886 79,729 298,930 425,546

Wages 297,525 358,887 1,939,752 2,596,164

Fringes 94,755 87,227 391,724 573,706

Operating supplies 158,493 275,786 1,284,868 1,719,148

General supplies 39,024 73,156 356,455 468,635

Tax and license 31,238 45,345 329,234 405,827

Insurance 40,743 60,480 316,131 417,354

Utilities 12,525 21,188 100,546 134,259

Depreciation 51,963 80,938 458,853 591,754

Equipment rents 322,098 689,313 2,655,641 3,667,052

Office equipment 5,523 15,472 45,125 66,120

Disposal of assets (2,516) (8,905) (30,218) (41,639)

Miscellaneous 13,800 27,606 73,231 114,637

Total expenses 1,112,057 1,806,233 8,220,271 11,138,562

Highway miles operated(thousands) 723,052 1,367,380 6,845,397 8,935,829

Ton-miles (thousands) 5,252,908 20,198,788 106,832,649 132,284,345

Cost per mile (dollars) 1.54 1.32 1.20 1.25

Cost per ton-mile (cents) 21.17 8.94 7.69 8.42

Average load (tons) 7.26 14.77 15.61 14.80

SOURCE: ATA (1995, Summary Tables III and V).


movements. To cope with these complexities, a number of researchers havedeveloped econometric cost estimation models.

Previous modeling efforts

Most econometric models are intended to measure changes in rail productivity overtime, as well as estimate the effects of mergers. Examples include Caves et al.(1980, 1981a, 1981b, 1981c) and Bereskin (1996). Models that estimate the nature ofeconomies of scale or density have been constructed by Spady (1979), Spady andFriedlaender (1976), Friedlaender and Spady (1980), Bereskin (1983), Barbera et al.(1987), and Lee and Baumel (1987).3 These authors generally conclude that the railindustry has become more productive over time. Of particular importance to ourwork, these modeling efforts have shown that rail costs are not linear in nature.

In a review of previous studies, Keaton (1990) reveals significant economies ofdensity in the general or mixed freight rail sector and conjectures that similareconomies of density may not exist in the case of unit trains (long trains carryingbulk cargo, such as grain). Keaton further suggests that some economies of densityare likely for intermodal trains. Several points are clear: the literature suggests thateconomies of scale and density exist in freight rail, and these economies probablyvary considerably among different types of rail operations.

Four rail scenarios

Because freight rail operations vary widely, a single aggregate value for private costper ton-mile would hold little meaning. To estimate private rail operating costs forrepresentative operational scenarios, we have developed cost models for four verydifferent types of freight trains.

• Heavy unit train. The train has 100 lightweight cars of 26 tons, andeach car carries 105 tons of cargo. The trip is 1,000 miles in length,with a 100 percent empty return. Power for the train consists of four3,000 brake horsepower (BHP) locomotives.

• Mixed freight train. Mixed cargo is carried in 90 cars averaging 32tons. The cargo averages 70 tons per car, and the trip length is 500miles, with a 45 percent empty return rate. Power for the train isprovided by three 3,000 BHP locomotives.

• Intermodal train. This train consists of 120 truck trailers riding on 120articulated spine cars. Trailers average 28 tons, including cargo, andspine cars weigh 14 tons. The trip length is 1,750 miles, and a fivepercent empty return rate is assumed. Power is supplied by three3,000 BHP locomotives.

3 Economies of scale result if unit costs are lower for larger railroad firms. Grimm and Harris

(1983, p. 275) point out that such economies are likely to be associated with theadministrative rather than operating functions of the firm. Economies of density result frommore frequent service on a given length of route, or from operating longer trains.

Private Costs 11

• Double-stack container train. The train consists of lightweight, five-well platform cars, with an average weight of 16 tons per well (80tons per car). Each well carries two containers with an averageweight of 28 tons, or 56 tons per well. There are 24 cars in the train(120 wells) carrying a total of 240 containers. A ten percent emptyreturn rate occurs. Power consists of four 3,000 BHP locomotives.

The four trains vary substantially in terms of basic configuration, power, trailing tonsof cargo, trip length, and empty return rates. While we assume the same accidentrates and noise impacts for all trains in Chapter 3, we vary emissions costs per ton-mile. Most important, the private cost per ton-mile varies among the four scenarios.Therefore, realistic and representative private cost estimates for different types offreight trains provide bases for comparing external costs with those experienced byrailroad companies.

An analysis of rail operating costs

As part of our research, we modeled operating costs of Class I railroads (those withannual gross operating revenues in excess of $50 million in 1978 dollars).4 To modelthese operating costs, we used a translog function (see Bereskin 1998). The functionhas four input prices: labor, materials and supplies, fuel, and other factors (using theAssociation of American Railroads index for other expenses). We also incorporatefour output measures: gross ton-miles, car-miles, train-miles, and locomotive-horsepower-miles. Data are for a 17-year period, 1978 through 1995. A total of 36firms are included in the analysis, but through mergers and bankruptcies only 11firms remained in 1995. Dummy variables are used as proxies for the changingrailroad structure due to changes in the number of firms.

Cost estimates are developed for each of the four stereotypical train types listedabove. These trains have very different operating parameters; our intent is to estimatethe costs of operating hypothetical but realistic train configurations. We havedeveloped two cost estimates for each train scenario, one with the operatingparameters averaged and one with the parameters weighted by gross ton-miles ofeach included railroad firm. Using both definitions enables us to examine costs astotal traffic and route density increase with both railroad firm size and volume. Weobserved sizable economies of size and density.

Our ton-mile operating cost estimates use averaged operating parameters for the fourrail scenarios and are presented in Table 2–2. Both the heavy unit train and themixed freight train scenarios result in ton-mile costs of approximately 1.2 cent. Theintermodal train cost per ton-mile is 2.68 cents, and the double-stack train costs 1.06cent. We compare these estimates with external costs in Chapter 5.

4 Class II railroads are those with annual gross operating revenues of between $10 and $50

million in 1978 dollars; Class III railroads have annual gross operating revenues of less than$10 million in 1978 dollars.


SUMMARY

Because TL carriers move over 90 percent of the intercity freight shipped by truck,we focus on this segment of the industry. The preponderant evidence suggests that inan unregulated environment, there are very limited economies of scale in TLtrucking, so using average cost data enables us to make reasonable estimates of per-ton-mile costs. Our primary interest is in general freight trucking because it accountsfor the vast majority of ton-miles in the TL trucking industry. In 1994, the averagecost per ton-mile for TL general freight trucking was 8.42 cents.

Within the rail freight sector there appear to be economies of scale, and costs ofdifferent types of service vary substantially. To provide representative estimates ofton-mile costs, we have developed four rail shipment scenarios and estimated costsfor each. One of the most valuable outputs of this analysis is an insight into thedegree of cost variance by type of train. Our estimates of operating costs range from1.06 cent per ton-mile for a particular configuration of container train to 2.68 centsfor an intermodal train.

In this chapter we estimate private trucking and freight rail ton-mile operating costsincurred by carriers. These estimates constitute the basis for comparisons withexternal costs developed in the next two chapters.

Table 2–2. Private operating costs of four railroad freight scenarios, 1994

Railroad scenario PowerCargo(tons)

Distance(miles)

Average costper ton-mile(1994 cents)

Heavy unit train 4 – 3,000 BHP locomotives 10,500 1,000 1.19

Mixed freight train 3 – 3,000 BHP locomotives 6,300 500 1.20

Intermodal train 3 – 3,000 BHP locomotives 3,360 1,750 2.68

Double-stack train 4 – 3,000 BHP locomotives 6,720 1,750 1.06

SOURCE: Research by Bereskin (1998).

Non-market Costs 13

CHAPTER 3NON-MARKET COSTS

To charge the full cost of transportation services, it is necessary to estimate themagnitude of social costs as accurately as possible. The difficulty is that few ofthese social costs can be assigned dollar amounts that will be widely embraced.Also, the effects of different transportation modes on the environment; infrastructure;other travelers; and the health, safety, and welfare of the general population is notfully understood. Keeping these limitations in mind, this chapter categorizes externalcosts of transportation and provides the best possible estimates of their magnitudes.

External costs can be subdivided into two general classes, those related to impactson other people who are not fully compensated and those related to unrecoveredgovernment expenditures. Subsidies result when a particular type of traveler does notpay the full cost of using public infrastructure.

The most important general categories of external costs arising from freight-carryingtransportation modes operating between cities are:

• accidents;

• emissions;

• noise impacts; and

• unrecovered costs associated with the provision, operation, andmaintenance of public facilities.

Although congestion is also a significant external cost of highway-basedtransportation, it is rarely a problem outside metropolitan areas. Given the inter-urban focus of our analysis, we therefore do not include congestion costs. The issueof unrecovered costs associated with public facilities is addressed in Chapter 4. Theother three primary categories of external costs germane to inter-urban freighttransportation are examined in the following sections.

ACCIDENTS

All transportation modes occasionally are involved in accidents and mishaps ofvarious sorts. When this occurs, people and their property often experience adverseoutcomes. The external cost due to accidents of a unit of transportation service (e.g.,a vehicle-mile or a trip) includes the uncompensated cost of deaths, injuries, andproperty damage that occur due to an additional trip by the mode in question.

It is not possible to provide completely accurate estimates of the marginal accidentcosts of trips by truck or rail. Inaccuracies stem from the effects of various types of


traffic on the accident rates of other travelers and non-traveling populations.Accident records can be examined to measure freight truck or train involvement inmotor vehicle accidents, but there is no systematic way to determine what role theyplayed in these accidents. If the primary cause of an accident is another vehicle, apedestrian, or conditions external to vehicles (e.g., severe weather), an involvedtruck or train may not have precipitated the incident. On the other hand, its presencemay have contributed to external costs experienced by other travelers or persons.

It is important to stress that fault is not at issue. Whether a truck or train involved inan accident was completely free of blame or whether it caused the accident isirrelevant to our analysis. Regarding the issue of fault, in a classic work, Vickery(1968) concludes:

(I)n most of the accidents with which we are concerned there aretwo or more parties involved, and the damage involved in theaccident could have been totally avoided if any party had acteddifferently, whether by driving less recklessly in the case of the“guilty” party, or by driving more defensively in the case of the“innocent” party, or by accomplishing the purpose in some way notinvolving the specific activity at all, as by travelling by train ratherthan automobile, or by living closer to one’s place of work, or evengiving up the object of the trip entirely… Systems which requirepayments by the actors only in the case of fault and only to theextent of the compensation received by others (even with theexpenses of adjudication and administration added) fail to give anadequate incentive for seeking out alternatives not involving theincreased risk of vehicular accident (pp. 466–467).

The real point is that the social cost would not have arisen had the particulartransportation service not been provided. Thus, a fatality or injury bears the samesocietal cost whether the affected person is an employee aboard a train or truck, anoccupant of another vehicle, or a pedestrian.

Our approach is to estimate total accident costs to society, per unit of serviceprovided, that result from accidents for each of the two transportation modes beingstudied. Costs to society consist of fatalities, personal injuries, and property damage.We provide data on the number of incidents and apply cost estimates to arrive attotal estimated costs. We then estimate the amount of compensation railroads andmotor carriers have provided. Subtracting this compensation from total societal costsdue to accidents yields external costs, the uncompensated accident costs that resultfrom motor carrier and railroad operations.

The cost of accidents

It is unpleasant to think of fatalities or personal injuries in monetary terms, but that iswhat must be done if one is to estimate the cost to society of accidents.Considerable work has been devoted to conceptual issues related to placing a valueon saving human lives and preventing personal injuries. Generally, the approach that

Non-market Costs 15

is becoming dominant is “willingness to pay.”5 According to this concept, the cost ofa particular type of accident is the amount people would pay to reduce the risk of ithappening.

To estimate willingness to pay for risk reduction, one observes market trade-offs inthe amount people pay for risk reduction versus other goods. Because some peoplewould be willing to pay more for a good than the asking price (that is, they enjoywhat economists refer to as consumer surplus), the amount that people pay for thegood is a lower-limit estimate of the value that they place on it.

For example, suppose we observe that four million people pay $100 each for a safetyenhancement on the new cars they buy. Further, suppose the buyers expect thisenhancement to reduce their chances of fatal injury by one in 4,000 over the periodthat they will be using the cars. As a group, the buyers expect their $400 millioninvestment to save 1,000 lives. Collectively, the buyers have demonstrated awillingness to pay $400,000 per life saved.

A report prepared for the Federal Highway Administration by the Urban Institute(Miller et al. 1991) summarized the results of numerous studies of the value of riskreduction. The values suggested by Miller et al. are widely used as estimates of theeconomic value of reducing the risks of motor vehicle crashes. Miller et al. expresstheir suggested values in 1988 dollars; in Table 3–1 we present their values in 1994dollars, having applied the Gross Domestic Product (GDP) deflator. These are thevalues used in our analysis.

Estimated external accident costs per ton-mile for freight trucks and trains follow.

Table 3–1. Cost of accidents (1994 dollars)

Accident type Per person Per accident

Fatal 2,903,782 3,304,027

Personal injury 56,255 84,455

Property damage 2,110 5,448

SOURCE: Miller et al. (1991), inflated to 1994 dollars.

Motor carriers

Evidence suggests that motor vehicle accident rates for fatal, personal injury, andproperty damage accidents increase with traffic volume up to a certain level oftraffic, about 7,000 vehicles per lane per day (Hall and Pendleton 1990). Forkenbrockand Foster (1997) have estimated the relationship between average daily traffic(ADT) per lane and accident rates per million vehicle-miles of travel (VMT). Usingsemi-logarithmic regression and data on 17,767 rural non-interstate highway 5 For discussions of the willingness to pay concept of value, see Viscusi (1993), National

Safety Council (1993), Jones-Lee (1989), and U.S. Office of Management and Budget (1991).


segments with ADT per lane ranging from 50 to 5,000, they found a significantpositive association between traffic volume and accident rate. For example, ahighway with 4,000 ADT per lane would have an accident rate 47.4 percent higherthan one with ADT per lane of 2,000 (p. 87).6 This suggests that the marginalaccident cost occasioned by one more vehicle operating on most roads and highwayswill exceed the average of those already on the roadway. Because of datalimitations, however, this analysis is based on average accident costs and thereforemay have a downward bias in terms of the marginal cost of a vehicle trip.

Accident costs. National accident data indicate that on a per-mile basis, theaccident costs of large trucks are significantly less than those of passenger cars.7

Table 3–2 shows that in 1994 large trucks had a fatal accident rate (per 100 millionVMT) nearly one-third greater than passenger cars, though injury and property-damage-only accident rates were considerably less for large trucks. Applying theaccident cost values from Table 3–1, we estimate total accident costs for passengercars and large trucks per 100 million VMT.

In 1994 TL general freight carriers transported an average of 14.80 tons of freight pervehicle-mile (ATA 1995, Summary Table III). Thus, the average cost to society ofaccidents related to general freight trucking was one cent per ton-mile.

Table 3–2. Accident rates and costsof passenger cars and large trucks, 1994 (per 100 million VMT)

Rate

Vehicle type FatalPersonalinjury

Propertydamage

Estimated cost(1994 dollars)

Passenger car 2.1 191 351 24,982,000

Large truck 2.7 56 211 14,800,000

SOURCE: Accidents rates from NHTSA (1996, Table 3).

Compensation. To estimate the per-ton-mile external cost of general freighttrucking, we determine the amount of compensation paid by affected truckingcompanies. When this amount is subtracted from the total cost to society, anestimate of the external cost results. Compensation by trucking companies has twoprincipal forms: 1) payment of workers’ compensation premiums and 2) payment ofpersonal liability and property damage insurance.

Like most other businesses, trucking companies purchase personal liability andproperty damage insurance. This insurance provides a means for compensating those

6 See also Lundy (1965) and Ceder and Livneh (1982). These authors also found a positive

association between traffic volumes and accident rates.7 The National Highway Traffic Safety Administration (NHTSA) defines a large truck as

having a gross vehicle weight greater than 10,000 pounds. Elsewhere in this report, we usea cut-off of 25,000 pounds.

Non-market Costs 17

who are injured, whose family member is killed, or whose property is damaged inaccidents involving a motor carrier’s vehicles. While almost all types of insuranceinvolve deductible payments by the insured business, these payments tend to bequite small relative to insurance premiums. Generally, corporate balance sheetsgroup premium and deductible loss payments.

As is true of other forms of insurance, personal liability and property damageinsurance costs are based on accident experience. This experience applies to theindustry of which a business is a part and to the specific experience of the individualbusiness. In 1994, general freight trucking companies paid $417,354,000 in personalliability and property damage insurance (ATA 1995, Summary Table III).

Each of the 50 states has a workers’ compensation law. While the laws varysomewhat, they provide for compensation of workers for employment-related injuriesand diseases without the need to determine fault. In 1988, 87 percent of all U.S.employees (91.3 million workers) were covered by workers’ compensation (U.S.Chamber of Commerce 1991, p. 1). Significantly, workers’ compensation actsgenerally exempt participating employers from damage suits. Premiums paid byemployers vary among industries, based on accident experience. In most states, anational organization, the National Council on Compensation Insurance, collectsaccident data and prepares rates.

Table 3–3 presents our estimate of the compensation paid by motor carriers. Theestimate of personal liability and property damage is specific to truckload (TL)general freight carriers. Our estimate of workers’ compensation was derived bycomputing 4.2 percent of total wages and salaries paid to employees of TL generalfreight trucking companies.8 According to the ATA (1995, Summary Table III), totalwages and salaries paid in 1994 by TL general freight trucking companies amountedto $3,021,710,000. Our estimate of workers’ compensation paid is 4.2 percent of thisamount, or $126,912,000.

Table 3–3. Compensation for accident costs paidby truckload general freight trucking companies, 1994 (dollars)

Source TotalPer 100

million VMT

Personal liability and property damage insurance 417,354,000 4,671,000

Workers’ compensation payments 126,912,000 1,420,000

Total 544,266,000 6,091,000

SOURCES: ATA (1995, Summary Table III); Bureau of Labor Statistics (1994, Table 4).

8 According to the Bureau of Labor Statistics (1994, Table 4), the 4.2 percent figure is the

average portion of total employee compensation for blue-collar workers accounted for byworkers’ compensation.


As Table 3–3 shows, we estimate that TL general freight trucking companies paid$6,091,000 in accident compensation per 100 million VMT. Using the ATA figure of14.80 tons transported per vehicle-mile, the compensation paid was 0.41 cent perton-mile.

To summarize, in 1994 TL general freight motor carriers were involved in accidentsthat cost society $14,800,000 per 100 million VMT. They paid compensation totaling$6,091,000 per 100 million VMT, leaving $8,709,000 in uncompensated accidentcosts. This accident externality equates to 0.59 cent per ton-mile.

Freight rail

Accident costs. Accidents involving freight trains fall into three primary categories:

• collisions at highway-rail grade crossings,

• persons struck by a train at other locations, and

• accidents involving the train alone.

The most frequent type of fatal accident is collisions at highway-rail grade crossings.Another major cause of fatal accidents is trains striking persons at locations otherthan grade crossings.9 No distinction is made here between trespassers and non-trespassers, though it should be noted that for all railroads taken together, trespassersaccount for the larger share of fatalities (55.9 percent). Most injuries, however,involve railroad employees on duty (81.6 percent) (Federal Railroad Administration1995, Table 14). In total there were 951 fatalities and 9,669 personal injurycasualties in 1994 arising from the operations of Class I freight railroads (FederalRailroad Administration 1995, Tables 38 and 39).10

Using the Miller et al. (1991) values expressed in 1994 dollars (see Table 3–1), thecosts to society of these fatal and personal injury casualties are $2,761,497,000 and$543,930,000, respectively (see Table 3–4). Property damage resulting from trainaccidents is difficult to estimate. One estimate of the value of property damage toother vehicles involved in crashes with trains at highway-rail grade crossings isprovided by the Bureau of Transportation Statistics (1997, Table 3–2) based onFederal Railroad Administration data. For 1994, the estimate is $18,553,000.11 Weassume that property damage for non-crossing rail accidents, other than that to trains,

9 For all freight railroads in 1994, highway-rail grade crossing crashes accounted for 50.4

percent of freight rail-related fatalities and 11.7 percent of the personal injuries. Personsstruck by a train at other locations accounted for 42.6 percent of the fatalities and 10.7percent of the personal injuries. Train accidents, per se, only accounted for 7.0 percent ofthe fatalities but fully 77.6 percent of the personal injuries (Federal Railroad Administration1995, Tables 13 and 14).

10 Both figures exclude Amtrak, a Class I passenger railroad, and other passenger fatalities andinjuries.

11 It is not possible to determine precisely what portion of this amount arose from operations ofClass I railroads, but we estimate the portion to be upwards of 90 percent.

Non-market Costs 19

is comparatively minor and ignore the costs of such damage. The total societal costof railroad accidents in 1994 dollars was about $3,323,980,000.

Compensation. To estimate compensation made by Class I railroads to victims ofaccidents involving trains, a different method is required than that used for motorcarriers. Railroad employees are compensated for on-the-job injuries through afederally mandated process based on tort claims in lieu of workers’ compensation.Also, railroads largely self-insure for personal liability and property damage.Specifically, they purchase insurance with a high deductible amount, $25 million ormore. Thus, in cases involving fatalities, injuries to non-employees, and propertydamage related to rail operations, most payments are made directly by the railroadthrough claims and suits.

When injured on the job, railroad workers (in contrast to employees of firms thatprovide transportation services using other modes) seek compensation under theFederal Employers’ Liability Act of 1908 (FELA). This act prescribes an approachbased on a tort process. To collect, a worker must demonstrate negligence on thepart of the employer, and awards are based on the degree of employee negligence.

Especially in recent years, the FELA process has become more similar to theworkers’ compensation system of most other industries (TRB 1994, p. 3). WhereasFELA was previously a more negligence-based, trial-driven tort system, now themajority of claims are settled without litigation or even legal representation.Conversely, within many other industries, litigation over workers’ compensationsystems has been increasing, though resolution rarely requires court action.

In general, the FELA process results in more sizable benefits to injured workers thandoes workers’ compensation (TRB 1994, p. 3).12 The amount injured workers will becompensated remains less certain, however, and legal fees and other transactionscosts still constitute a larger portion of FELA settlements than is the case withworkers’ compensation (TRB 1994, pp. 4–5). According to the Association ofAmerican Railroads (nd), legal and administrative costs constitute on average 31

12 According to the Transportation Research Board (1994, p. 9), the rail industry would be more

competitive with other modes if a system like workers’ compensation were adopted, but theeffect would probably be modest.

Table 3–4. Costs of accidentsinvolving Class I freight rail, 1994

Accident type Amount (dollars)

Fatal 2,761,497,000

Personal injury 543,930,000

Property damageto other vehicles 18,553,000

Total 3,323,980,000


percent of FELA payments. As Table 3–5 shows, FELA compensation to injuredrailroad workers totaled $1.113 billion in 1994.

For other claims against them, U.S. railroads essentially self-insure. Typically, theirpersonal liability and property damage insurance has extremely high deductiblefeatures, often $25 to $100 million per event. Unlike motor carriers, then, railroadsgenerally compensate in the form of paid claims rather than insurance premiums.

Railroads report two general categories of claims: rail crossing accidents and otherincidents. In each case, the amount paid may be the result of negotiation orlitigation. In 1994, Class I railroads paid $97 million in claims for accidents at railcrossings and another $53 million in other accident claims (see Table 3–5). Addingtogether rail liability, property damage claims, and FELA judgments, Class I railroadspaid a total of $1.263 billion in compensation in 1994 for accidents involving freighttrains. Although we are able to include the payout level for property damage as aresult of train accidents at rail crossings, there are no similar data on payment levelsfor other accidents such as train derailments. Still, it is doubtful that the amount paidout for property damage resulting from other accidents is large relative to the totalaccident compensation paid by Class I railroads (shown in Table 3–5).

Table 3–5. Compensation for accidentcosts paid by Class I freight railroads, 1994

Source Amount (dollars)

FELA (railroad employees) 1,113,000,000

Claims for rail crossing accidents 97,000,000

Claims for other accidents 53,000,000

Total 1,263,000,000

SOURCE: Correspondence from the Association of American RailroadsLaw Department dated March 29, 1996.

In summary, Class I freight railroads were involved in accidents that cost society atotal of $3,323,980,000 in 1994, and they paid a total of $1,263,000,000 in variouskinds of compensation for accidents. The net uncompensated accident cost of freightrail operations in 1994 was therefore $2,060,980,000. Dividing this figure by the1,200,701,000,000 Class I rail ton-miles in 1994 (Bureau of Transportation Statistics1997, Table 1–9) results in an uncompensated cost of 0.17 cent per ton-mile.

EMISSIONS

Vehicle-generated air pollution is an external cost because the transportation userswho are producing it are not the only ones who are affected. Two general categoriesof emissions from internal combustion engines are important to our analysis ofexternal costs: air pollution and greenhouse gases. We begin by estimating the costs

Non-market Costs 21

of air pollution generated by freight truck and rail, then consider the costs ofgreenhouse gas emissions.

Nature and magnitude of air pollution

The use of internal combustion engines to transport passengers and freight is a majorcause of air pollution. In 1994, highway vehicles accounted for 62.3 percent of allcarbon monoxide (CO) emissions and 31.9 percent of all emissions of nitrogen oxides(NOX) in the United States; for railroads the percentages are 0.2 for CO and 4.0 forNOX (Bureau of Transportation Statistics 1996b, Tables 7–2 and 7–4). Significantamounts of hydrocarbons (HC) also were generated by the transportation sector.

It is not easy to estimate the external costs of emissions generated by specific typesof transportation users, and two basic elements must be understood before full-costpricing can be implemented. The first element is the amount of air pollutionassociated with a unit of travel by different types of vehicles operating underdifferent conditions. The second is the dollar value of damage to human health andother things of value—animals, crop yields, buildings and structures, and scenicviews. This analysis incorporates the best available estimates of these two elements.

Transportation sources account for about 45 percent of the Environmental ProtectionAgency’s (EPA’s) six criteria pollutants (TRB 1995, p. 39), briefly defined in Table3–6. For diesel engines, the primary emission type is NOX (nitrogen oxides),followed by CO (carbon monoxide), PM10 (particulate matter under ten microns inaerodynamic diameter), SO2 (sulfur dioxide), and VOC (volatile organic compounds,mainly hydrocarbons) (TRB 1995, p. 45).

Some pollutants are much more localized in nature than others. For example, in ruralareas CO emissions are likely to be concentrated along heavily traveled corridors buttoo dispersed to adversely affect the health or life quality of rural populations (seeForkenbrock and Schweitzer 1997a, Chapter 3). SO2, on the other hand, leads to theformation of acid rain which can adversely affect vegetation, buildings, and humansover a large area. Similarly, VOC and NOX react to produce ground-level ozone(O3), which leads to regional smog production.

Costs of air pollution

Placing dollar values on units of air pollution is at best problematic. For one thing,another unit of pollution bears a higher social cost in a locale where that form ofpollution is already relatively high. In places where pollution is high enough toapproach maximum allowable levels (beyond which health is threatened), furtherpollution is especially costly. Because this analysis is concerned with freightmovements between U.S. cities, however, the preponderance of truck or railoperations occurs in rural areas where ambient pollution levels are generally low.Especially for more localized pollutants, using high values for the cost of pollutionwould thus tend to overstate the true costs to society of air pollution stemming fromfreight movements using either rail or truck.


Table 3–6. Types of air pollution*

Pollution type Description

Carbonmonoxide

CO Odorless gas formed by incomplete combustion of motor fuels. Thehigher the fuel-to-air ratio for an engine, the more CO is produced.Fuel-rich operations occur when an engine is cold or is operatingunder a greater load (steep hills or higher speeds). The currentstandards are no more than nine parts per million (ppm), not to exceedeight hours per year or 35 ppm for not more than one hour per year.Two-thirds of all CO emissions come from the transportation sector.

Nitrogenoxides

NOX These gases, formed by high-temperature chemical processes thatoccur during the combustion of fuel in an internal-combustion engine,lead to the formation of smog and contribute to the formation of acidrain. The current standard is an annual mean of 0.05 ppm. About 40percent of all NOX emissions come from the transportation sector.

Volatileorganiccompounds

VOC Chemical compounds that leave the engine through the exhaustsystem or crankcase and from the fuel system through evaporation.Ground-level ozone (O3), which leads to smog formation, is producedthrough a chemical reaction between VOC and NOX. Thetransportation sector accounts for about half of all VOC.

Particulatematter

PM 10 Inhalable particular matter (less than 10 microns in diameter)emanating from the exhausts of diesel-powered vehicles (a minorsource) and road dust, along with more pervasive sources outside thetransportation sector. Currently, the standards are no more than 150micrograms per cubic meter of air for a 24-hour period or 50micrograms per cubic meter on average for a year. Vehicle emissionsper se account for less than one percent of all PM10 emissions, but dustraised by vehicle movement (so-called “fugitive dust”) is a much moreimportant source.

Sulfur oxides SO X Compounds that lead to the formation of particulate matter andsulfuric acid (leading to acid rain). Of greatest interest here is sulfurdioxide (SO2). Diesel-powered vehicles emit SO 2 because of the highsulfur content of diesel fuel, but coal-fired electrical utilities are amuch greater source of SO2 emissions. The current standards are nomore than 0.14 ppm for a 24-hour period or 0.03 ppm over the courseof a year. Approximately one-eighth of all SO2 emissions are producedby the transportation sector.

Lead Pb Transportation sector emissions of this pollutant are insignificantbecause of the adoption of lead-free motor fuels; lead is therefore notconsidered in our analysis.

*Fractions of the respective pollutants accounted for by the transportation sector are from U.S.EPA (1994a).

SOURCES: Current pollution standards are from Code of Federal Regulations (1995) and U.S.EPA (1997).

Very limited research has been published that assigns actual dollar values to thecosts of vehicle-generated air pollution. Small and Kazimi (1995, Table 5) provideestimates of the costs of VOC, NOX, SOX, and PM10 in the Los Angeles region, aplace with comparatively high pollution levels. Haling and Cohen (1995) usedresults of work by National Economic Research Associates (NERA 1993) to estimatethe costs of VOC, NOX, SOX, and PM10 for 2,233 rural U.S. counties in various states(there are 3,048 rural counties in the United States). Using these estimates, weaveraged the costs for the 2,233 rural counties. Table 3–7 shows the results in terms

Non-market Costs 23

of dollars per ton of pollution emitted, using a value of $2.9 million per statisticallife (see Table 3–1).13

NERA bases its emission costs on existing knowledge of adverse effects on health,materials, agriculture, and aesthetic quality brought about by higher ambientconcentrations of several air pollutants. Two types of health effects were studied: 1)small changes in the probability of premature death and 2) nonfatal effects rangingfrom minor irritations to more serious ailments that require medical treatment.

Based on epidemiology literature, NERA concluded that changes in the probability ofpremature death due to air pollution are primarily related to PM10 emissions (seeLippmann and Thurston 1996, Schwartz et al. 1996, and Pope et al. 1995). Nonfatalhealth effects were estimated using epidemiological and clinical studies of ill effectsrelated to pollutant exposure (see Schwartz and Morris 1995 and Thurston et al.1994). Damage to materials was valuated using a damage cost function, andagricultural crop damage costs were derived from U.S. EPA-sponsored field tests. Thevalue of aesthetic damage caused by pollution concentrations was estimated usingloss of visibility (expressed in visual range miles) as a proxy for overall aestheticdamage; a hedonic price model was applied to infer the deleterious effect of reducedvisibility on housing values.

Even though the NERA study is the most salient research effort our reviewuncovered, NERA’s results (upon which we base Tables 3–9, 3–11, and 3–12) cannotbe regarded as definitive. While an impressive level of research has been conductedon the health effects and other adverse impacts of air pollution, precise risk factorsand the economic value that should be assigned to them are not yet a matter ofconsensus.

It is important to note that the cost estimates in Table 3–7 are low compared tostudies that have focused on urban areas. For example, Wang and Santini (1995,Table 5) use data for 17 metropolitan areas to estimate dollar values per ton of NOX;at a minimum, their estimates are almost four times larger than those in Table 3–7.

13 NERA had used a value of $4.0 million per statistical life.

Table 3–7. Average air pollutant costs for2,233 rural counties (1994 dollars)

Pollutant Cost per ton

VOC 385

NOX 213

SO X 263

PM 10 3,943

SOURCE: Derived from Haling andCohen (1995).


The SOX cost estimates in Wang and Santini are over eight times higher, but theirPM10 cost estimates are a third less than those in our table.

We now apply the resultant pollution costs to freight trucks and rail. Estimates ofemission levels for the two transportation modes are presented, and cost estimatesper ton-mile are developed.

Motor carriers. Emission factors for various truck configurations are shown in Table3–8. Of special relevance here are heavy duty diesel trucks, which are used fortruckload (TL) general freight transportation between cities. In particular, we areinterested in emission factors for such trucks at 55 mph, a common cruising speed.

Table 3–8. Emission rates for selected truck configurations (grams per mile)

Truck type and speed VOC NOX SO X PM 10

Light-duty gasoline

35 mph 2.53 2.31 0.038 0.055

40 mph 2.33 2.35 0.038 0.055

45 mph 2.19 2.39 0.038 0.055

50 mph 2.11 2.61 0.038 0.055

55 mph 2.09 3.12 0.038 0.055

Light-duty diesel

35 mph 0.71 1.72 0.122 0.380

40 mph 0.64 1.77 0.122 0.380

45 mph 0.59 1.88 0.122 0.380

50 mph 0.55 2.07 0.122 0.380

55 mph 0.52 2.37 0.122 0.380

Heavy-duty diesel

35 mph 1.74 14.76 0.576 1.527

40 mph 1.56 15.16 0.576 1.527

45 mph 1.43 16.12 0.576 1.527

50 mph 1.34 17.77 0.576 1.527

55 mph 1.28 20.29 0.576 1.527

NOTES:

NO X and VOC factors are developed from EPA’s MOBILE5a model using the followingassumptions: operating modes 20.6/27.3/20.6, an ambient temperature of 87.5 with a minimumof 72 and maximum of 92 degrees Fahrenheit, default diesel sale shares, low altitude with nooxygenated fuels, and reformulated gasoline (RFG) included (see U.S. EPA 1994b).

Particulate emission factors are developed from EPA’s PART5 model with the followingassumptions: low altitude region, no improvement and maintenance program, particle sizecutoff of ten microns, and no transient RFG. The particulate emission factors include PM 1Oexhaust, brake, and tire sources (see U.S. EPA 1995).

SOURCE: Cambridge Systematics (1995, Table 4.2).

Non-market Costs 25

Multiplying the emission factors from the shaded area in Table 3–8 by theappropriate costs in Table 3–7, we obtain estimates of the costs of air pollution pervehicle-mile. Table 3–9 shows that diesel-powered heavy trucks transporting 14.80tons on average in rural areas produce air pollution costs of 0.08 cent per ton-mile.

Table 3–9. Emission rates and costs for freight trucks

EmissionRate (grams

per ton-mile)*Cost per gram

(cents)†Cost per ton-mile

(cents)*

VOC 0.086 0.042 0.004

NOX 1.371 0.023 0.032

SOX 0.039 0.029 0.001

PM10 0.103 0.435 0.045

Total 0.082

*Based on an average of 14.80 tons in cargo transported per vehicle-mile.† Derived from Table 3–7.

We should note that our emission rates are somewhat conservative. Harrison et al.(1992, Table 3) citing Weaver (1988) suggest emission rates two to three timesgreater: VOC, 0.292; NOX, 2.585; and PM10, 0.359 grams per ton-mile. Emissionrates are influenced by assumptions regarding the condition of the trucking fleet,operating circumstances such as topography, and the amount of stop-and-go driving.Our estimated air pollution cost is also somewhat less than that found in or derivedfrom several other studies. Modifying the results of an analysis by Small and Kazimi(1995, Table 8) to estimate 1994 fleet average emission rates and using a value of$2.9 million for a statistical life (see Table 3–1 of this monograph), we obtain a totalcost of pollution per vehicle-mile for a heavy-duty diesel truck of $0.287. Dividingthis figure by 14.80 tons per vehicle-mile yields a per-ton-mile cost of 1.9 cent. It isimportant to stress that this estimate is derived from an analysis of the Los Angelesregion, which has substantially greater ambient air pollution levels than rural areas.

An even higher estimate of 2.8 cents per ton-mile for air pollution generated byheavy diesel-powered trucks is derived from results published by the EuropeanCommission (1996, Annex 10). Because this estimate is based on poorly documentedassumptions, direct comparison between it and other estimates is difficult.

Freight rail. In the case of freight rail, we estimate emission rates per ton-mile ofcargo for the four general types of freight trains discussed in Chapter 2. With adaptedemission rates from Barth and Tadi (1996), Table 3–10 presents VOC, NOX, andPM10 emissions per ton-mile for the four types of trains.14 As discussed in Chapter 2,the four scenarios vary considerably in terms of locomotive power and trailingtonnage of cargo. Consequently, the emissions rates also vary among the scenarios.

14 Barth and Tadi (1996) do not include SOX. It is an inconsequential pollutant of freight rail.


Table 3–10. Emission rates for four typesof freight trains (grams per ton-mile)*

Type of train VOC NOX PM 10

Heavy unit train 0.003 0.257 0.006

Mixed freight train 0.004 0.322 0.008

Intermodal train 0.007 0.603 0.015

Double-stack train 0.005 0.400 0.010

* In each scenario the average train speed is 45 mph, and the throttlesetting is notch 8 (full power).

SOURCE: Adapted from Barth and Tadi (1996, Table 1).

In Table 3–11, we apply the estimated emission costs per ton from Table 3–7 tocalculate the air pollution cost per ton-mile of the four general types of freight trains.On a ton-mile basis, the total costs of air pollution for any of the general types offreight trains are very small, not more than 0.02 cent.

Table 3–11. Emission costs of four typesof freight trains (1994 cents per ton-mile)

Type of train VOC NOX PM10 Total

Heavy unit train 0.0001 0.006 0.003 0.009

Mixed freight train 0.0002 0.008 0.003 0.011

Intermodal train 0.0003 0.014 0.006 0.020

Double-stack train 0.0002 0.009 0.004 0.013

From this analysis, it is fair to conclude that the external costs of air pollutiongenerated by freight rail operating in rural areas are very small. A higher estimate ofthese costs is made by the European Commission (1996, Table A.6). The EuropeanCommission estimates air pollution costs associated with shipping 1,000 tons onekilometer to be 1.8 European currency units (ECU). Converting this figure to U.S.dollars per ton-mile, an estimate of 0.22 cent emerges. It is unclear what type offreight train, trailing tonnage, or speed were used in the analysis.

Applying emission rates from Blevins and Gibson (1991, Table 7) results in estimatesvery close to those in Table 3–11. While these authors’ rail scenarios differ from ourssomewhat, they arrive at similar emission rates for VOC, NOX and PM10.

Greenhouse gas emissions

While not technically air pollution, greenhouse gas emissions constitute a threat tosociety by contributing to global climate change. Carbon dioxide (CO2) is by far themost prominent greenhouse gas released by human activity, accounting for about 85percent of total emissions weighted by global warming potential (Bureau ofTransportation Statistics 1996b, p. 144). Within the United States in 1994, the

Non-market Costs 27

transportation sector accounted for about 32 percent of all CO2 emissions (U.S.Department of Energy 1995, Table 5).

Although there is uncertainty as to the likely climatic impacts of the greenhouseeffect, the role of the transportation sector in the production of greenhouse gases isclear. CO2 is a byproduct of any engine that burns carbon-based fossil fuels. Theamount of CO2 released per unit of transportation service (i.e., per ton-mile) isdirectly related to the energy efficiency of the mode providing that service. Onegallon of diesel fuel releases 22.8 pounds of CO2 (FHWA 1997b, p. I–5).

Researchers have yet to reach anything close to a consensus regarding the cost tosociety of releasing CO2 into the atmosphere. Chernick and Caverhill (1991, Table11–1) document estimates of up to $24.95 but as low as $2.27 (1994 dollars) per ton,while NERA (1993, Table 63) suggests $3.56 (1994 dollars). Perhaps one of the morecredible estimates is that of a National Research Council study (NRC 1991) thatsuggests costs in the range of $10 to $20 per ton of CO2 emitted. In our analysis, weuse the lower figure of $10.

Because the amount of CO2 emitted is directly proportional to the quantity of dieselfuel burned, one can estimate CO2 emissions and apply the cost of $10 per ton. Fortrucking, if we use a fuel efficiency of 5.2 miles per gallon (Bureau of the Census1995, Table 11; see also FHWA 1993, Table VM–1) and an average payload of 14.80tons per vehicle-mile, the cost to society of CO2 emissions per ton-mile shipped bytruck is 0.15 cent per ton-mile. CO2 emissions per ton-mile for three trainconfigurations (mixed freight, intermodal, and double-stack) are estimated by Blevinsand Gibson (1991, Table 7). Using their estimated emission rates, the societal cost ofCO2 emissions per ton-mile shipped is 0.02 cent per ton-mile. Table 3–12 presentsthe rates and costs of CO2 emissions for both modes.

Table 3–12. Rates and external costs of carbondioxide emissions, truck and rail, 1994

Freight modeEmission rate perton-mile (grams)

Cost per ton-mile(1994 cents)

Heavy truck 134.4 0.15

Mixed freight rail 18.6 0.02

Intermodal rail 17.0 0.02

Double-stack rail 15.4 0.02

SOURCE: Rail emission rates are from Blevins and Gibson (1991, Table 7).

Conclusions

To estimate the social costs of air pollutants and greenhouse gases emitted by freighttruck and rail, we have applied the best available data on rates and unit costs. Ineach case where disparate estimates exist, we used an estimate that is relatively


conservative. We must stress, however, that epidemiologists have yet to pinpoint themarginal increases in health problems associated with small increases in pollutionlevels. We do know that metropolitan areas with higher concentrations of airpollutants are virtually certain to experience greater negative health impacts as aresult of an increase in these pollutants than is the case in most rural areas whereconcentrations are much lower. Because this analysis pertains to intercity freightmovements, using relatively low pollutant costs seems prudent. The CO2 costestimates, however, would not vary by location of occurrence because greenhousegas emissions are a global problem.

Although we have disaggregated freight-carrying vehicles as much as possible, thereis by necessity considerable averaging in the estimates contained in this analysis.Per ton-mile pollution levels will be higher in hilly terrain or when freight trains ortrucks are transporting fewer net tons per mile of travel, for example.

NOISE

Transportation is a major source of noise pollution, which can be defined asunwanted or detrimental sound. While transportation-generated noise affects fewerpeople in rural areas, noise is a negative externality wherever train or truck trafficoccurs.

Measuring noise impacts

The general measure of sound is the decibel (dB). Measuring truck and rail trafficnoise involves an adjustment or weighting of high- and low-pitched sounds toapproximate human perception of these sounds. Decibels of sound adjusted toapproximate human perception are called “A-weighted levels” (dBA). On an A-weighted decibel scale, zero denotes the faintest sound humans can hear. The scaleis logarithmic. To most people, for example, noise at 60 dBA (an air conditioningunit) sounds twice as loud as noise at 50 dBA (a clothes dryer) and noise at 70 dBA(a pickup truck) is perceived to be four times as loud as noise at 50 dBA.

Surveys have indicated that people often think of traffic noise as the most disruptiveindoor problem caused by nearby highways (Williams and McCrae 1995, p. 80). Theamount of disruption caused depends on the volume, speed, and composition of thetraffic. Composition is the most important factor: one combination truck produces asmuch noise at 55 mph as 28 automobiles at the same speed (FHWA 1992, pp. 4–5).A combination truck heard from 50 feet away produces a noise level of 90 dBA.

Noise decreases at more than a linear rate with distance (Ishiyana et al. 1991, pp.69–70). Traffic noise tends to dwindle away from the road until it reachesbackground levels at about 1,000 feet from a highway source (Hokanson et al. 1981,p. 7). Table 3–13 shows an example of the relationship between traffic volume,noise, and distance from the road. Conditions other than distance (e.g., topography,trees, and buildings) can influence noise levels, however, so the figures in Table 3–13 should be considered approximations.

Non-market Costs 29

Table 3–13. Decibel levels based ontraffic volume and distance from the road

Distance (feet)

Average daily traffic 55 dBA 65 dBA

Up to 7,999 404 57

8,000–27,999 736 159

28,000–47,999 970 209

Greater than 48,000 1,339 289

SOURCE: Hokanson et al. (1981, p. 17).

Effects of noise

Depending on the individual person, noise can cause sleep disturbance,communication interference, and general annoyance. Communication interferenceoccurs when nearby traffic masks normal conversation, causing people to strain tohear and be heard. “Annoyance” generally describes physical and psychologicalstress. Table 3–14 summarizes the results of increasing noise levels, includingannoyance and communication interference.

Table 3–14. Effects of noise levels

Noise level (dBA) Effect

55–64 Annoyance

65–69 Communication interference

70–79 Muscles and glands react

Above 80 Changed motor coordination

SOURCE: Bureau of Transportation Statistics (1994, p. 169).

Because people vary greatly in the degree to which they tolerate noise, an absolutechange in noise level is not necessarily a predictor of annoyance. In the case ofhighway noise, for example, researchers have found that individual differences innoise tolerance explain more variance in comfort level than do differences in noiseitself. Some individuals show high annoyance at 60 dBA, while others remainunconcerned at 80 dBA (Langdon 1985, pp. 163–167). The intensity, duration,predictability, and controllability of noise are related to the negative impacts ofnoise on each individual (Llewellyn 1981, pp. 192–196). Even if noise levels are lowon average, intermittent noise can be bothersome, especially for people who need toconcentrate, rest, or maintain tranquillity. Schools, nursing homes, hospitals, andchurches are places where intermittent noise is particularly intrusive.

In addition to (or perhaps because of) the psychological effects of noise, which arevery difficult to monetize, noise tends to have an adverse effect on residential


property values. Studies of this effect have associated the noise level in decibelsabove an established noise threshold with average changes in property values.Average property value changes per decibel increase can then be tied to the noisegenerated by a particular type of vehicle operating at various distances from theproperty.

Modeling truck-generated noise

The Federal Highway Administration (FHWA) has developed a simple noiseemission model pertaining to heavy trucks (FHWA 1979, p. 7; see also Rudder et al.1979, p. 4). The model takes into account the speed, number of axles, and weight ofthe truck to predict the peak noise level. Noise dispersion models like the FHWASTAMINA model (FHWA 1982) or derivatives such as the Minnesota Department ofTransportation MINNOISE model (Mn/DOT 1991) are applied to estimate noiselevels at specified distances from a highway. These models represent a noisegradient that falls at a logarithmic rate (see Hokanson et al. 1981, p. 9).

In conjunction with geographic information systems (GIS), dispersion models can beused to define noise contours, or lines of equal noise levels along transportationcorridors. The number of people living within areas where the amount of noiseexceeds state or federal standards can then be estimated. To make these estimates,average population per housing unit and average number of housing units per squaremile for general categories of land use are taken into account. One such category is“rural character,” and it is the most relevant to this analysis of intercity traffic.

Hokanson et al. (1981, p. 10) estimate that near highways (within the distance rangewhere noise levels exceed 55 dBA) sparse rural development has an average of 0.006housing units per acre, and five housing units on average are found in areas of denserural development. Using these values, Haling and Cohen (1996, Table 4) estimatethat 0.51 housing units are impacted per rural highway system mile (or, presumably,per mile of rail trackage) in areas of dense rural development. Where ruraldevelopment is sparse, they estimate that essentially no housing units are subjectedto traffic noises high enough to lower property values.

In the case of housing units in densely settled rural areas, Haling and Cohen (1996,Table 7) estimate that heavy traffic moving at 55 mph produces a noise cost of about0.24 cent per VMT. This estimate is derived from the work of Nelson (1978) whoconcluded that housing unit values fall by 0.4 percent per dBA increase in theconstant equivalent noise level. Haling and Cohen (Table 9) estimate a per-vehicle-mile noise cost of 4.7 to 6.9 cents for a combination truck, depending on the truck’soperating weight. For housing units in sparsely settled rural areas, they estimate nonoise cost from vehicles. Nationally far more miles of rural highway run throughsparsely settled areas than those that are densely developed. If we assume that 90percent of all intercity freight truck VMT occur in sparsely settled rural areas and tenpercent of the VMT are in densely settled rural areas, an approximate noise cost per

Non-market Costs 31

freight truck mile is 0.58 cent. On an average ton-mile basis (14.80-ton load), theestimate is 0.04 cent.

Rail noise impact

Far less has been written about noise generated by freight rail operating in rural areasthan about heavy trucks.15 Most of the limited literature pertains to high speedpassenger rail in Europe, rather than slower moving freight trains. Fath et al. (1974)have measured noise levels at 450 feet from track centerline for trains of varyinglength and trailing weight. Noise levels ranged from 70 to 90 dBA. Hanson et al.(1991) have estimated the incremental change in freight rail noise with differenttypes and conditions of track and train wheels. Among the limited noise of freightrail estimates available are Planco (1990) who places the cost at 0.22 cent per ton-mile (adjusted to 1994 dollars) and Diekmann (1990) whose figure is 0.20 cent perton-mile (also in 1994 dollars). Both authors studied Germany where ruraldevelopment is generally denser than in the U.S., so their estimates are likely to berelatively high compared to ours.16

In general, the literature suggests that a given level of noise produced by a freighttrain is usually perceived as less annoying than noise produced by vehicle traffic ona highway. The so-called “Green Book” of the Commission of the EuropeanCommunities (OECD 1992) implies that the cost of road traffic noise is over sixtimes greater than noise from freight rail. Similar results are reported by Rothengatter(1989), who found that the ratio of people “annoyed” by road noise compared to railnoise is 3.4 to one; the ratio of people “highly annoyed” is 6.4 to one.

We are unaware of compelling evidence that in rural areas of the United Stateshighway noise is actually more objectionable than freight rail noise. Accordingly,we use the same value of 0.04 cent per ton-mile for both freight modes. In the caseof either mode, the social cost of noise per ton-mile of transportation service in ruralareas is very small.

CONCLUSIONS

In this chapter we have examined available evidence on the magnitude of accident,emission, and noise external costs generated by freight truck and rail operatingbetween cities. We have not considered certain costs that occur within metropolitanareas, such as traffic congestion. Also, values for air pollution and noise are muchlower than would be the case in urban areas.

15 A good discussion of the technical aspects of train noise is found in Nelson (1987, Section

1.4.2).16 For comparison, the estimates per ton-mile of truck freight are 0.32 cents for Planco and

0.18 for Diekmann. These figures are between four and eight times larger than the value of0.04 cents for freight trucks we use.


Throughout this monograph we have used ton-miles as the unit of analysis. As wementioned earlier, it is not particularly meaningful to make direct comparisonsbetween truck and rail because of the generally very different natures of these twomodes’ operations. The more relevant comparisons are between external costs andprivate costs for each mode; these comparisons are made in Chapter 5.

In Table 3–15 we summarize the results of the three analyses contained in thischapter. The total external cost for heavy freight trucks operating between cities is0.86 cent per ton-mile. For freight rail the per-ton-mile external cost totals 0.24 to0.25 cent.

Table 3–15. Summary of external costs of truckand rail freight (1994 cents per ton-mile)

AccidentsAir

pollution*Greenhouse

gases Noise Total

General freight truck 0.59 0.08 0.15 0.04 0.86

Heavy unit train 0.17 0.01 0.02 0.04 0.24

Mixed freight train 0.17 0.01 0.02 0.04 0.24

Intermodal train 0.17 0.02 0.02 0.04 0.25

Double-stack train 0.17 0.01 0.02 0.04 0.24

* Totals from Tables 3–9 and 3–11 rounded to two decimal places.

Government Subsidies 33

CHAPTER 4GOVERNMENT SUBSIDIES

Whenever a transportation user operates a vehicle on a public facility, he or sheoccasions some level of cost. This cost arises from wear and tear on the roadwayover and above what would occur without travel by that user, as well as otherpublicly borne costs. When the user pays less than the cost occasioned, a de factosubsidy occurs. Depending on how the facility is financed, this subsidy amounts to atransfer from other users of the facility or from society more generally.

In this chapter we review what is known about the costs occasioned and the usercharges paid by freight trucks. Because freight rail generally operates on its ownfacilities, the issue of user subsidies does not emerge in the case of rail service.17

COSTS OCCASIONED BY VEHICLES

The costs occasioned on a given trip depend both on the type of vehicle making thetrip and on characteristics of the roadway on which the vehicle is operating. In thisdiscussion we focus on infrastructure costs that ultimately result in the need forexpenditures by the governmental agency with jurisdictional responsibility.

Highway cost allocation studies

Both the federal government and most states periodically conduct highway costallocation studies primarily to estimate the responsibility of different vehicle classesfor highway costs and the user fees paid by the respective vehicle classes.Comparing the two estimates, one is able to assess which vehicle classes, onbalance, overpay and which underpay relative to other classes.

Because other external costs were addressed earlier, this discussion concerns“agency costs” (costs borne by governmental agencies that provide the relevanthighway facilities). Federal agency costs largely pertain to the primary costcategories listed in Table 4–1 (pavements, bridges, and other elements of highwayinfrastructure that wear out and must be refurbished or replaced). When estimatingagency costs and how they should be distributed among vehicle classes, costallocation studies consider the functional classes of the highways on which eachvehicle class operates. The per-mile cost occasioned by a heavy vehicle may bequite small on newly reconstructed interstate highways, for example, but very highon lower standard roads.

17 According to Chapman and Martland (1997, Exhibit 7), railroads spend slightly more than $8

billion annually on track and right-of-way maintenance.


Table 4–1. Estimated distribution of federalhighway obligations in year 2000

Category Improvement typePercentof total

New capacity New construction 10.8

Reconstruction with added lanes 3.4

Major widening 6.8

Subtotal 21.0

System enhancement Safety/TSM* 9.4

Environmental 1.9

Other projects 4.1

Subtotal 15.4

System preservation 3R preservation† 26.7

Minor widening 1.8

Major bridge 1.6

Bridge replacement 7.8

Minor bridge 4.4

Subtotal 42.3

Transit 12.4

Other 8.9

Total 100.0

*Transportation system management.

†Reconstruction, resurfacing, and rehabilitation.

SOURCE: FHWA (1997a, Table III–7).

Current methods of charging vehicles for their use of the highway system (primarilymotor fuel taxes and registration fees) do not enable federal or state user charges tovary according to the particular highway on which the travel occurs. Thus, averagecost responsibilities for different vehicle classes must be derived. Such costresponsibilities should be viewed with some caution because they generally will notreflect the actual agency cost occasioned by a given vehicle on a specific trip. Inother words, user charges based on average cost responsibilities for a given vehicleclass bring about some level of cross-subsidization among vehicles within that class.

The 1997 Federal Highway Cost Allocation Study

In August 1997, the Federal Highway Administration released the 1997 FederalHighway Cost Allocation Study (Federal HCAS), the first such study in 15 years. Theprimary objective of the study was to evaluate the equity and efficiency of current


highway user charges. For purposes of our research, estimates of the magnitude ofgovernment subsidies to freight trucks are based on the equity ratios (user chargespaid divided by cost responsibilities) produced by the Federal HCAS.

Several points should be kept in mind when applying the Federal HCAS equityratios. First, HCAS derived two estimates of cost responsibilities for various classesof vehicles. One is based on the traditional cost-occasioned approach which isconcerned with the costs created by vehicles of different sizes and weights in termsof the design, construction, rehabilitation, and maintenance of highways. The otherestimate approximates the marginal social cost which does not consider agencyexpenditures, but instead considers the economic cost to society of increments ofhighway use by each vehicle class.

The economic costs in the second estimate include both infrastructure costs that leadto agency expenditures and some environmental and other social costs. Because wealready have developed estimates of environmental and safety costs, it wouldconstitute double counting to include all of the economic costs of the second FederalHCAS estimation method. Additionally, the second approach includes congestioncosts which are ignored in this study of freight transportation between urban areas. Inthis analysis, we apply the traditional cost-occasioned approach.

A second important point in applying the findings of the Federal HCAS is that thestudy includes estimates that pertain to highway cost allocation for all levels ofgovernment, as well as for the federal aid highway system alone. An advantage ofconsidering equity ratios for all levels of government is illustrated by a simpleexample. The federal user charge for a particular vehicle class may be too low (i.e.,the vehicle class is underpaying), but at the state level the same vehicle class maybe overpaying. Taken together, the two levels of government may have an equityratio for the vehicle class that is close to 1.0. Such an equity ratio implies that thevehicle class is paying society fully for the costs the class occasions.

The estimates for all levels of government are by necessity highly aggregated andtherefore not indicative of the circumstances in any given state or locality.Nonetheless, these estimates constitute the best picture of the overall governmentsubsidy, if any, received by operators of the respective vehicle classes. A summaryof the various cost categories for all levels of government to be allocated to highwayusers is presented in Table 4–2.

DO FREIGHT TRUCKS UNDERPAY?

Table 4–3 shows a forecast of highway user charge payments at all levels ofgovernment in the year 2000.18 Of the $97.7 billion in total imposts, $27.2 billion(27.8 percent) will be collected via federal user charges, $68.5 billion (70.1 percent)

18 The all-levels-of-government analysis in the Federal HCAS is based entirely on forecasts for

the year 2000.


Table 4–2. Highway obligations or expenditures for 1994and year 2000 at all levels of government ($ millions)

Obligation or expenditure

1994(1994

dollars)

Year 2000(year 2000

dollars)

Annualgrowth

rate (%)

Federal obligationsDirect from the Highway Trust Fund (HTF) 1,394 1,819 4.53Direct from other sources 340 413 3.30Total direct federal obligations 1,734 2,232 4.29Transfers to states from HTF 16,916 21,644 4.19Transfers to states from other funds 569 691 3.30Transfers to local governments from HTF 259 331 4.19Transfers to local governments from other funds 440 535 3.30Mass transit expenditures from HTF 2,304 3,380 6.60Total federal aid to state and local 20,488 26,581 4.43Total federal obligations from HTF 20,873 27,174 4.49Total federal obligations from other funds 1,349 1,639 3.30

Total federal obligations 22,222 28,813 4.42

State expendituresCapital outlays 32,059 40,868 4.13Maintenance 7,152 8,961 3.83Traffic services 2,984 4,211 5.91Administration and research 4,847 6,841 5.91Debt service 4,318 6,094 5.91Law enforcement and safety 4,209 5,940 5.91

Subtotal 55,569 72,915 4.63Federal transfers (17,485) (22,330) 4.16Grants-in-aid to local governments 8,838 11,596 4.63Local government transfers (1,381) (1,710) 3.63Motor vehicle administration 2,667 3,499 4.63Motor fuel administration 244 320 4.63Mass transit expenditures 1,796 2,356 4.63

Total state expenditures 50,248 66,641 4.82

Local expendituresCapital outlays 9,231 10,630 2.38Maintenance 11,796 14,085 3.00Traffic services 1,771 2,115 3.00Debt service 3,859 5,325 5.51Administration, law enforcement, and safety 6,133 8,463 5.51

Subtotal 32,790 40,618 3.63Federal transfers (699) (866) 3.64Grants-in-aid to local governments (8,838) (11,594) 4.63Transfers to states 1,381 1,710 3.63

Total local expenditures 24,634 29,868 3.26

Total obligations and expenditures (all levels) 97,104 125,322 4.34

SOURCE: FHWA (1997a, Table III–8).


Table 4–3. Year 2000 highway user charge payments by vehicle classat all levels of government (year 2000 dollars in millions)

Vehicle class/registered weight Federal State Local Total

Passenger vehicles

Autos 11,576 34,520 1,164 47,264

Pickups and vans 5,812 16,260 479 22,554

Buses 20 311 6 337

All passenger vehicles 17,408 51,090 1,649 70,155

Single unit trucks

≤ 25,000 pounds 1,500 3,831 84 5,415

25,001–50,000 pounds 611 1,802 58 2,471

> 50,001 pounds 487 924 31 1,442

All single unit trucks 2,598 6,558 173 9,329

Combination trucks

≤ 50,000 pounds 306 560 15 881

50,001–70,000 pounds 504 902 22 1,427

70,001–75,000 pounds 370 548 14 932

75,001–80,000 pounds 5,521 7,651 145 13,317

> 80,000 pounds 468 1,156 12 1,636

All combination trucks 7,169 10,810 208 18,195

All trucks 9,766 17,370 380 27,522

All vehicles 27,174 68,470 2,029 97,677

SOURCE: FHWA (1997a, Table IV–11).

will come from state-level user charges, and $2.0 billion (2.1 percent) will begenerated by local user charges. The table shows the amounts forecast to becollected from 11 classes of vehicles, as well as aggregations of these changes.

Table 4–4 displays the cost responsibilities of the same 11 vehicle classes forecast inthe year 2000, as estimated in the 1997 Federal HCAS. The total cost responsibilityfor the year 2000 is $125.3 billion, or about $27.6 billion more than the FHWAforecasts will be collected at all levels of government. This means that as a group,those who use the roads and highways provided by all levels of government will notpay user charges equal to the total costs they occasion.

Among highway users, the 11 classes of vehicles vary considerably in terms of thefraction of their cost responsibilities defrayed by the user charges they pay. Dividingthe user charge payments in Table 4–3 by the cost responsibilities in Table 4–4,ratios of payments to cost responsibilities are derived (see Table 4–5).


Table 4–4. Year 2000 federal cost responsibility for all levelsof government by vehicle class (year 2000 dollars in millions)

Vehicle class/registered weight Federal State Local Total

Passenger vehicles

Autos 12,405 35,988 15,791 64,184

Pickups and vans 4,770 13,678 6,328 24,777

Buses 221 383 268 871

All passenger vehicles 17,396 50,049 22,387 89,832

Single unit trucks

≤ 25,000 pounds 1,074 1,755 886 3,715

25,001–50,000 pounds 981 1,867 1,349 4,197

> 50,001 pounds 1,098 1,929 1,212 4,239

All single unit trucks 3,153 5,551 3,447 12,151

Combination trucks

≤ 50,000 pounds 222 325 149 696

50,001–70,000 pounds 528 722 306 1,555

70,001–75,000 pounds 408 517 178 1,103

75,001–80,000 pounds 6,329 8,353 2,950 17,632

> 80,000 pounds 778 1,125 450 2,353

All combination trucks 8,264 11,042 4,032 23,338

All trucks 11,417 16,593 7,479 35,490

All vehicles 28,813 66,642 29,866 125,322

SOURCE: FHWA (1997a, Table V–21).

Several important conclusions can be reached from the figures in Table 4–5. At alllevels of government, the aggregate ratios of user charge payments to costresponsibilities for passenger vehicles (0.781), single unit trucks (0.768), andcombination trucks (0.780) do not vary much. This suggests that while highway usersas a group are only paying about 77.9 percent of the costs they occasion, no majorcategory of vehicles is overpaying or underpaying substantially relative to allhighway users.

We should stress that the per-mile subsidy varies somewhat among the classes oftrucks. As Table 4–6 shows, combination trucks with registered weights of between70,000 and 80,000 pounds underpay by 4.9 cents per vehicle-mile, whilecombination trucks registered between 50,000 and 70,000 pounds underpay by only1.2 cents per vehicle-mile. Overall, combination trucks on average underpay 4.4cents per vehicle-mile. The figures in Table 4–6 include both LTL and TL truckingand types of service other than general freight. ATA data, however, show that TL


general freight trucking has a higher average payload (and hence weight) than allcarriers taken together: 14.80 versus 14.39 tons per vehicle-mile (ATA 1995,Summary Table I). Thus, the overall figure of 4.4 cents may slightly understate theamount by which this type of trucking service underpays.

Using an average of 14.80 tons of freight transported per vehicle-mile, combinationtrucks will underpay by 0.30 cent per ton-mile in year 2000 dollars. Deflating thisfigure to 1994 dollars, unrecovered costs associated with the provision, operation,and maintenance of public facilities at all levels of government equal 0.25 cent perton-mile.

Table 4–5. Ratios of year 2000 user charge paymentsto allocated costs for all levels of government

Vehicle class/registered weight Federal State

Federaland state Local

All levels ofgovernment

Passenger vehicles

Autos 0.993 0.959 0.953 0.074 0.736

Pickups and vans 1.218 1.189 1.196 0.076 0.910

Buses 0.109 0.812 0.548 0.022 0.387

All passenger vehicles 1.001 1.021 1.016 0.074 0.781

Single unit trucks

≤ 25,000 pounds 1.397 2.183 1.884 0.095 1.458

25,001–50,000 pounds 0.623 0.965 0.847 0.043 0.589

> 50,001 pounds 0.444 0.479 0.466 0.026 0.340

All single unit trucks 0.824 1.181 1.052 0.050 0.768

Combination trucks

≤ 50,000 pounds 1.378 1.723 1.583 0.101 1.266

50,001–70,000 pounds 0.955 1.249 1.125 0.073 0.918

70,001–75,000 pounds 0.907 1.060 0.992 0.079 0.845

75,001–80,000 pounds 0.872 0.916 0.897 0.049 0.755

> 80,000 pounds 0.602 1.028 0.853 0.027 0.695

All combination trucks 0.867 0.979 0.931 0.052 0.780

All trucks 0.855 1.047 0.969 0.051 0.775

All vehicles 0.943 1.027 1.001 0.068 0.779

NOTE: These ratios are based on total revenues and expenditures nationwide and are likely tovary from ratios for individual states and local governments. Federal ratios includeobligations not financed by the Highway Trust Fund.


Table 4–6. Underpayment of highway user charges by vehicle classfor all levels of government, year 2000 (year 2000 dollars)

Vehicle class

User chargeunderpayment($ millions)

Vehicle-milestraveled(millions)

Underpayment(cents per VMT)

Passenger vehicles

Autos 16,920 1,818,461 0.9

Pickups and vans 2,223 669,198 0.3

Buses 534 7,397 7.2

All passenger vehicles 19,677 2,495,056 0.8

Single unit trucks 2,822 83,100 3.4

Combination trucks

≤ 50,000 pounds (185) 6,743 (2.7)

50,001–70,000 pounds 128 10,683 1.2

70,001–80,000 pounds 4,486 92,102 4.9

> 80,000 pounds 717 23,587 3.0

All combination trucks 5,143 115,689 4.4

All trucks 7,968 198,789 4.0

All vehicles 27,645 2,693,845 1.0

SOURCE: Derived from FHWA (1997a, Tables II–10, IV–11, and V–21).

CONCLUSIONS

Applying results of the 1997 Federal Highway Cost Allocation Study, we havecompared the cost responsibilities of trucks with different configurations and trucksregistered to operate at various weights. Lighter combination trucks, especially thoseunder 50,000 pounds, underpay the least; heavier single-unit trucks underpay themost. The relatively commonplace five-axle combination truck weighing 70,000 to80,000 pounds underpays almost five cents per vehicle-mile. Overall, combinationtrucks underpay 4.4 cents per vehicle-mile, or 0.25 cent per ton-mile.

Internalizing External Costs 41

CHAPTER 5INTERNALIZING EXTERNAL COSTS

In this concluding chapter, we compare the external costs of truck and rail freighttransportation operating between U.S. cities with the relevant private costs. We alsoexplore practical methods for internalizing external costs.

PRIVATE AND EXTERNAL COSTS

We have estimated private costs of intercity freight transportation in Chapter 2,external costs in Chapter 3, and de facto subsidies in Chapter 4. Some assumptionsare necessary to carry out the analyses underlying these estimates, and ourconclusions depend on a careful interpretation of earlier research by others. We firstrecap the roles of assumptions and antecedent research.

Role of assumptions

Because our analysis focuses on intercity freight movements, we generally ignoresuch urban impacts as traffic congestion. Although intercity trucks travel from urbanfactories and warehouses to rural highways, by far the greatest portion of the journeyoccurs on rural highways where congestion is very rarely a significant problem.

Likewise, we apply comparatively low values for the cost of air pollution with therationale that urban areas typically experience relatively higher ambient pollutionlevels than rural areas. The health effects of additional pollutants in already pollutedareas are bound to be more severe than in cleaner rural areas (Small and Kazimi1995, p. 8). Even though we use comparatively low values for air pollution costsbecause of our rural focus, we cannot be certain that the societal cost is reflectedprecisely. To date, definitive values of the social cost of various air pollutants havenot been published.

The largest externality for both rail and truck transportation is safety-related. Severalkey assumptions influence our results in the area of safety, including the valuesassigned to fatal and personal injury accidents. There is only limited consensus onthe appropriate values to use for transportation-related mishaps. In a survey of all 50state departments of transportation conducted in 1993, we found they use an averagevalue of $1.2 million for fatalities and $41,000 for injuries (Forkenbrock et al. 1994,p. 18). FHWA, however, is urging use of the higher values recommended by Milleret al. (1991); these are the values we use in this analysis, inflated to 1994 levels.We should note that studies of the health effects of environmental pollution often usehigher values. Hall et al. (1992, p. 815) found that recommended values range from$4.0 million to $9.2 million for the worth to society of saving a statistical life. We


have used fairly conservative values for the economic costs of accidents, recognizingthat such valuation is by nature normative and imprecise.

Assumptions such as those just reviewed certainly affect our estimates of private andexternal costs. Because we have opted to use relatively conservative values for keyparameters throughout our analysis, the magnitudes of the accident, air pollution,greenhouse gas, and noise externalities generally are comparatively low. Values forpublic facility costs presented in Chapter 4 are also conservative.

Comparison of private and external costs

Table 5–1 summarizes our best estimates of private costs, external costs, and usercharge underpayment for general freight trucking and four freight rail scenarios. Forfreight trucks, the total external cost and user charge underpayment is 1.11 cent perton-mile. For freight rail, external costs are substantially smaller: 0.24 to 0.25 centper ton-mile. Because the average private cost is so much smaller for rail than fortrucking, however, external costs for rail often are larger relative to private costs.

Table 5–1. Private and external costs of truckand rail freight (1994 cents per ton-mile)

Private cost(1)

External cost(2)

User chargeunderpayment

(3)(2) + (3) as

percent of (1)

General freight truck 8.42 0.86 0.25 13.2

Heavy unit train 1.19 0.24 20.2

Mixed freight train 1.20 0.24 20.0

Intermodal train 2.68 0.25 9.3

Double-stack train 1.06 0.24 22.6

Implications for trucking

Table 5–1 indicates that for general freight trucking operations between cities, theper-ton-mile cost would increase by 13.2 percent if all external costs and usersubsidies were internalized. User charges would need to be increased aboutthreefold to reflect all external costs and user subsidies in the costs of this mode.

To illustrate the implications of such an increase in user charges for TL generalfreight carriers, consider a truck that travels 76,700 miles per year (the averagederived from ATA 1995, Summary Table III). The approximate private operating costof the truck is $95,900 per year ($1.25 per vehicle-mile, see Table 2–1), and theoperator pays 39.8 cents per gallon in fuel taxes (the 50-state average).19 To 19 The federal diesel fuel tax is 20 cents per gallon (excluding 4.3 cents for deficit reduction).

According to the Congressional Budget Office (1998, p. 56), the weighted average state-level diesel fuel tax is 19.8 cents per gallon.


internalize the external cost of $12,660 (13.2 percent of $95,900), the per-gallon taxwould need to be increased to $1.26.20

The foregoing example is illustrative, not by any means definitive. As Forkenbrockand Schweitzer (1997b) point out, motor fuel taxes are not able to take into accountthe specific circumstances in which a vehicle is operating. Trucks that operate onlow-standard roadways, in metropolitan areas with comparatively high levels of airpollution, or where population densities are higher (so more people are exposed totraffic-related noise) generate greater social costs. These authors suggest usingintelligent transportation system (ITS) technology to fine-tune road user charges toparticular operating circumstances.

Applying the ITS technology discussed by Forkenbrock and Schweitzer, it would bepossible to assess road user charges based on road segments and taking into accountvehicle weight and configuration, roadway and bridge design, environmental factors,and state-specific user charge levels. Figure 5–1 depicts the approach schematically.The important point is that freight truck user charges, unlike motor fuel taxes, wouldbe tailored to reflect the costs actually occasioned on a particular trip.21

Of relevance to this research, external costs associated with air pollution, noise, andaccidents could be factored into road segment classifications. A road segmentlocated in an area with relatively high ambient pollution (e.g., a federal air qualitynon-attainment area) could have a higher air pollution charge. Similarly, proximity toresidences, schools, hospitals, or other activities adversely affected by vehicle noisecould be taken into account in the charge for traveling on a road segment. Based onaccident data, higher user charges could be levied on road segments where theprobability is higher that a truck will be involved in an accident.

Building comprehensive cost factors into road segment classifications would helpaccomplish three important public policy objectives: to compensate society for thesocial costs occasioned by a heavy vehicle on a particular trip, eliminate cross-subsidies among heavy vehicles, and encourage vehicle operators to travel onhighways where costs would be lower.

20 A fuel efficiency of 5.2 miles per gallon is assumed (Bureau of the Census 1995, Table 11;

see also FHWA 1993, Table VM–1).21 In brief, the approach discussed by Forkenbrock and Schweitzer (1997b) would involve a

small computer on board freight trucks. The computer would receive vehicle weightinformation from an on-board weighing device that would update weight data to thecomputer when the cargo doors are closed. A global positioning system (GPS) receiverwould provide location information, which would be reconciled with stored geographic datathat establish the location of road segments. Also coded in the computer would be thevehicle identification and configuration. It would thus be possible to generate an electronicrecord of road use, taking into account vehicle weight and configuration, roadcharacteristics, and other location-related factors. Research on this concept continues.


Global Positioning System• Latitude/longitude coordinates

of travel route

Vehicle On-boardWeight Indicator

Vehicle GISComputer Software

Vehicle On-board Computer Database• Vehicle number• Date of travel• Vehicle type identifier• Stored road segment coordinates• Road segment classifications• State/local government unit of segment• Axle weight

Road Use Summary• Vehicle identification• Date• Travel in miles by: — road classification — state/locality of travel — vehicle weight

Data Processing Center• Prepares billing statements• Maintains auditable summaries

of road use

Figure 5–1. ITS approach to charging heavy vehicles for road use

SOURCE: Forkenbrock and Schweitzer (1997b, Figure 2).

Implications for freight rail

Adding external costs to the private costs currently experienced by railroadcompanies would require a very different approach than for trucking. Becauserailroads provide their own infrastructure, they do not often pay user charges to


governmental agencies for facility use. Fortunately, in Table 3–15 we see that theexternal costs per ton-mile do not vary appreciably among different train types. Thus,a reasonably simple charge of about one-quarter of a cent per ton-mile could beadded to rail shipments.

It is true that the largest type of external costs, accident costs, is not directly relatedto ton-miles transported. Rather, accident costs are probably more dependent ontrain-miles operated. The number of miles operated affects accident risk exposure bypedestrians and operators of motor vehicles, with whom trains may conflict.

Specifically, a ton-mile charge based on the external costs of accidents would to adegree discriminate against a railroad company whose trains tend to operaterelatively few miles and are short in length but carry very heavy cargo. Conversely,a railroad company operating relatively long, lightweight trains that travel manymiles would be favored by a tax on ton-miles.

Quite likely, however, the basic economics of freight transportation make itimpractical to operate trains that carry predominately bulky, lightweight cargo. Railis most competitive in markets where the cargo is relatively heavy and the distancesare long. To the extent that the preponderance of rail traffic fits this profile,inequities arising from using ton-miles as a basis for internalizing external costswould not be serious.

CONCLUSIONS

Though important to the U.S. economy, freight trucking and rail create adverseimpacts. These impacts are referred to as external costs because they are not borneby those who generate them. Placing an appropriate dollar value on external costs isvital to the process of internalizing them; that is, requiring those who generate thesecosts to compensate society in an amount equal to the external costs.

Internalizing external costs makes it possible to return to society an amount equal tothe costs one imposes; it also gives a clear signal of the actual full cost of anactivity, so that consumption decisions can be made on the basis of this cost. In thecase of transportation, including external costs generally will lead to some reductionin consumption of transportation services, such as by locating producing facilitiescloser to markets. Another impact may be greater use of a competing mode that hasless total cost, once external costs are included.

Our analysis has focused on freight rail and TL general freight trucking between U.S.cities. This accounts for a very large share of the total ton-miles of transportation,and it enables us to make two important simplifications in the analysis. We are ableto ignore congestion costs because very little congestion exists in rural areas.Additionally, we are able to use single values for the costs of various air pollutants.Ambient pollution levels vary among metropolitan areas, so the social costs offurther vehicle-generated pollution will differ as well.


Four main types of external costs have been included in our analysis: accidents;emissions; noise; and unrecovered costs associated with the provision, operation, andmaintenance of public facilities. In each case we have used conservative values forthe various key parameters, so that our estimates of external costs constitute baselinevalues. Increasing the values assigned to preventing a fatality, reducing traffic noise,or improving air quality will lead to concomitantly higher external cost figures.

Our final estimates of external costs for intercity TL general freight trucking and railfreight transportation imply that these costs are substantial. For general freighttruckload trucking, the external cost is 1.11 cent per ton-mile. This figure is equal to13.2 percent of the private operating cost of that transportation mode. In the case offreight rail, the per-ton-mile external cost is only one-third as large as for trucking.Because the private cost (direct cost to the transportation provider) is much lower forrail (1.06 to 2.68 cents per ton-mile), rail external costs constitute larger amountsrelative to private costs. The range of external costs compared to private costs is 9.3to 22.6 percent.

The conclusion of this research is thus that even when using conservative values forexternal costs, these costs are sizable enough to warrant concern. External costsaffect the well-being of society and should be fully included in the decision-makingprocess of how much service by each transportation mode should be consumed. Ourresearch has sought to provide reasonable estimates of the amounts by whichintercity truck and rail transportation costs should be increased to include externalcosts. We hope that these estimates will help facilitate enlightened public andprivate sector decision making.

References 47

REFERENCES

Abacus Technology Corporation. 1991. Rail Versus Truck Fuel Efficiency: The

Relative Fuel Efficiency of Truck Competitive Rail Freight and Truck

Operations Compared in a Range of Corridors. Report prepared by AbacusTechnology Corporation. Washington, DC: U.S. Department of Commerce,National Technical Information Service.

American Trucking Associations (ATA). 1995. 1994 Motor Carrier Annual Report:

Financial and Operating Statistics. Alexandria, VA.

Association of American Railroads. 1994. Analysis of Class I Railroads 1994.Department of Economics and Finance. Washington, DC.

Association of American Railroads. 1996. Correspondence regarding railroadaccident claims. Law Department (March 29). Washington, DC.

Association of American Railroads. nd. Tort Abuse and the Rail Industry: The

Facts About FELA. Washington, DC.

Barbera, Anthony, Curtis M. Grimm, Kent A. Phillips, and Leslie J. Selzer. 1987.“Railroad Cost Structure—Revisited.” Journal of the Transportation Research

Forum, Vol. 28, No. 1, pp. 237–244.

Barth, Matthew, and Ramakrishna Tadi. 1996. “An Emissions Comparison BetweenTruck and Rail: A Case Study of the California I–40.” Paper presented at theTransportation Research Board 75th Annual Meeting, Washington, DC.

Bereskin, C. Gregory. 1983. “A Bi-level Model of United States Railroad Costs.”Unpublished dissertation. Columbia, MO: University of Missouri—Columbia.

Bereskin, C. Gregory. 1996. “Econometric Estimation of Post-Deregulation RailwayProductivity Growth.” Transportation Journal, Vol. 35, No. 4 (Summer),pp. 34–43.

Bereskin, C. Gregory. 1998. “Economic Estimation of Costs for Specific RailroadFreight.” Unpublished paper. Davenport, IA: St. Ambrose College, Departmentof Economics.

Blevins, W. G., and A. W. Gibson. 1991. “Comparison of Emissions and Energy Usefor Truck and Rail.” Paper presented at the Annual Conference of theTransportation Association of Canada. Winnipeg, Manitoba (September).


Bureau of the Census. 1995. 1992 Census of Transportation, Truck Inventory and

Use Survey. TC 92–T–52. Washington, DC: U.S. Department of Commerce.

Bureau of Labor Statistics. 1994. News. USDL:94–290. Washington, DC: U.S.Department of Labor (June 16).

Bureau of Transportation Statistics. 1994. Transportation Statistics Annual Report

1994. U.S. Department of Transportation. Washington, DC: U.S. GovernmentPrinting Office.

Bureau of Transportation Statistics. 1996a. National Transportation Statistics 1996.

U.S. Department of Transportation. Washington, DC: U.S. Government PrintingOffice.

Bureau of Transportation Statistics. 1996b. National Transportation Statistics

Annual Report 1996: Transportation and the Environment. Washington, DC:U.S. Government Printing Office.

Bureau of Transportation Statistics. 1997. National Transportation Statistics 1997.

U.S. Department of Transportation. Washington, DC: U.S. Government PrintingOffice.

Button, Kenneth J. 1993. Transport Economics, 2nd Ed. Cambridge, UK: UniversityPress.

Cambridge Systematics. 1995. Incorporation of External Cost Considerations in

Highway Cost Allocation. Report prepared by Cambridge Systematics, Inc.Washington, DC: U.S. Department of Transportation, Federal HighwayAdministration.

Caves, Douglas W., Laurites R. Christensen, and Joseph A. Swanson. 1980.“Productivity in U.S. Railroads, 1951–1974.” Bell Journal of Economics Vol.11, No. 1 (Spring), pp. 166–181.

Caves, Douglas W., Laurites R. Christensen, and Joseph A. Swanson. 1981a.“Economic Performance in Regulated and Unregulated Environments: AComparison of U.S. and Canadian Railroads.” Quarterly Journal of Economics,Vol. 96, No. 4 (November), pp. 559–581.

Caves, Douglas W., Laurites R. Christensen, and Joseph A. Swanson. 1981b. “TheHigh Cost of Regulating U.S. Railroads.” Regulation, Vol. 5, No. 1 (January/February), pp. 41–46.

Caves, Douglas W., Laurites R. Christensen, and Joseph A. Swanson. 1981c.“Productivity Growth, Scale Economies and Capacity Utilization in U.S.Railroads, 1955–1974.” American Economic Review, Vol. 71, No. 5(December), pp. 994–1002.

References 49

Ceder, A., and M. Livneh. 1982. “Relationship Between Accidents and HourlyTraffic Flow—I and II.” Accident Analysis and Prevention, Vol. 14, No. 1,pp. 19–44.

Chapman, Jeffrey D., and Carl D. Martland. 1997. “Railroad Track Productivity: AHistorical Perspective.” Transportation Quarterly, Vol. 51, No. 3 (Summer),pp. 105–118.

Chernick, Paul L., and Emily J. Caverhill. 1991. “Valuation of EnvironmentalExternalities in Energy Conservation Planning.” In Energy Efficiency and the

Environment: Forging the Link. Edward Vine, Drury Crawley, and PaulCentolella, eds. Washington, DC: American Council for an Energy-EfficientEconomy in cooperation with the University-wide Energy Research Group,University of California, pp. 215–228.

Code of Federal Regulations. 1995. Vol. 40, Section 50.6–50.10.

Congressional Budget Office. 1998. Innovative Financing of Highways: An

Analysis of Proposals. U.S. Congress. Washington, DC: Government PrintingOffice.

Diekmann, A. 1990. “Nutzen und Kosten des Automobils.” Internationales

Verkehrswesen, November–December.

European Commission. 1996. Towards Fair and Efficient Pricing in Transport:

Policy Options for Internalizing the External Costs of Transport in the

European Union. Luxemborg, Belgium: Office for Official Publications of theEuropean Communities.

Fath, J. M., D. S. Bloomquist, J. M. Heinen, and M. Tarica. 1974. Measurements of

Railroad Noise—Line Operation, Yard Boundaries, and Retarders. NBSIR 74–488. Washington, DC: U.S. Environmental Protection Agency, Office of NoiseAbatement and Control.

Federal Highway Administration (FHWA). 1979. Statistical Analysis of FHWA

Traffic Noise Data. FHWA–RD–78–64. Washington, DC: U.S. Department ofTransportation.

Federal Highway Administration (FHWA). 1982. Noise Barrier Cost Reduction

Procedure: STAMINA 2.0/OPTIMA User Manual. Washington, DC: U.S.Department of Transportation.

Federal Highway Administration (FHWA). 1992. Highway Traffic Noise.Washington, DC: Government Printing Office.

Federal Highway Administration (FHWA). 1993. Highway Statistics 1992.Washington, DC: U.S. Government Printing Office.


Federal Highway Administration (FHWA). 1997a. 1997 Federal Highway Cost

Allocation Study. U.S. Department of Transportation. Washington, DC:Government Printing Office.

Federal Highway Administration (FHWA). 1997b. “Federal Highway Cost AllocationStudy Draft Report Appendix I.” Washington, DC.

Federal Railroad Administration. 1995. Accident/Incident Bulletin No. 163.Washington, DC: U.S. Department of Transportation.

Forkenbrock, David J., Norman S. J. Foster, and Thomas F. Pogue. 1994. Safety and

Highway Investment. Iowa City, IA: University of Iowa, Public Policy Center.

Forkenbrock, David J., and Lisa A. Schweitzer. 1997a. Environmental Justice and

Transportation Investment Policy. Iowa City, IA: University of Iowa, PublicPolicy Center.

Forkenbrock, David J., and Lisa A. Schweitzer. 1997b. “Intelligent TransportationSystems and Highway Finance in the 21st Century.” In Proceedings of the

National Conference on Transportation Finance for the 21st Century.Washington, DC: Transportation Research Board, pp. 73–82.

Forkenbrock, David J., and Norman S. J. Foster. 1997. “Accident Cost Saving andHighway Attributes.” Transportation, Vol. 24, No. 1 (February), pp. 79–100.

Friedlaender, Ann F., and Richard H. Spady. 1980. “Economic Costs and the UniformRailroad Costing System.” Conference on Railroad Costing Procedures: InterstateCommerce Commission (June).

Greene, David L., and Donald W. Jones. 1997. “The Full Costs and Benefits ofTransportation: Conceptual and Theoretical Issues.” In The Full Costs and

Benefits of Transportation. David L. Greene, Donald W. Jones, and Mark A.Delucchi, eds. Berlin: Springer-Verlag, pp. 1–26.

Grimm, Curtis M., and Robert G. Harris. 1983. “Structural Economies of the RailFreight Industry: Concepts, Evidence, and Merger Policy Implications.”Transportation Research, Vol. 17A, No. 4 (July), pp. 271–281.

Haling, Daniel, and Harry Cohen. 1995. “Air Quality Cost Analysis Spreadsheet.”Data provided by the authors. Washington, DC: Cambridge Systematics, Inc.

Haling, Daniel, and Harry Cohen. 1996. “Residential Noise Damage Costs Causedby Motor Vehicles.” Transportation Research Record 1559, pp. 84–94.

Hall, J. W., and O. J. Pendleton. 1990. “Rural Accident Rate Variations with TrafficVolume.” Transportation Research Record 1281, pp. 62–70.

References 51

Hall, Jane V., Arthur M. Viner, Michael T. Kleinman, Frederick W. Lurmann, VictorBrajer, and Steven D. Colome. 1992. “Valuing the Health Benefits of CleanAir.” Science, Vol. 255, No. 5046, pp. 812–817.

Hanson, Carl E., Hugh J. Saurenman, and David A. Towers. 1991. “RailTransportation Noise and Vibration.” In Handbook of Acoustical

Measurements and Noise Control, Cyril M. Harris, ed. New York, NY:McGraw-Hill, Inc., pp. 46.1–46.23.

Harrison, Robert, Michael T. McNerney, and Mark A. Euritt. 1992. “DeterminingTruck System Costs for the Pennsylvania Interstate 80 Corridor.” Transportation

Research Record 1359, pp. 68–73.

Hokanson, Barry, Martin Minkoff, Steve Nichols, and Susan Cowart. 1981. Measures

of Noise Cost Attributable to Motor Vehicle Travel. Iowa City, IA: Universityof Iowa, Institute of Urban and Regional Research.

Ishiyana, Takeshi, Kazumasa Tateisha, and Toshiro Arai. 1991. “An Analysis ofTraffic Noise Propagation Around Main Roads in Tokyo.” Noise Control

Engineering Journal, Vol. 36, No. 2, pp. 65–72.

Jones-Lee, M. W. 1989. The Economics of Safety and Physical Risk. New York,NY: Basil Blackwell.

Keaton, Mark H. 1990. “Economies of Density and Service Levels on U.S. Railroads:An Experimental Analysis.” The Logistics and Transportation Review, Vol. 26,No. 3, pp. 211–227.

Langdon, F. L. 1985. “Noise Annoyance.” In The Noise Handbook. W. Tempest,ed. London, England: The Academic Press, pp. 143–176.

Lee, Tenpao, and C. Phillip Baumel. 1987. “The Cost Structure of the U.S. RailroadIndustry Under Deregulation.” Journal of the Transportation Research Forum,Vol. 28, No. 1, pp. 245–253.

Lippmann, Morton, and George D. Thurston. 1996. “Sulfate Concentrations as anIndicator of Ambient Particulate Matter Air Pollution for Health RiskEvaluations.” Journal of Exposure and Environmental Epidemiology, Vol. 6,No. 2, pp. 123–146.

Llewellyn, Lynn. 1981. “The Social Cost of Urban Transportation.” In Transportation

and Behavior. Irwin Altman, Joachim F. Wohlwill, and Peter B. Everett, eds.New York, NY: Plenum Press, pp. 162–202.

Lundy, R. A. 1965. “Effect of Traffic Volumes and Number of Lanes on FreewayAccident Rates.” Highway Research Record 99, pp. 138–147.


McMullen, B. Starr, and Linda R. Stanley. 1988. “The Impact of Deregulation on theProduction Structure of the Motor Carrier Industry.” Economic Inquiry, Vol. 26,No. 2 (April), pp. 299–317.

Miller, T., J. Viner, S. Rossman, N. Pindus, W. Gellert, J. Douglass, A. Dillingham,and G. Blomquist. 1991. The Cost of Highway Crashes. Report prepared by theUrban Institute. FHWA–RD–91–055. Washington, DC: U.S. Department ofTransportation, Federal Highway Administration.

Minnesota Department of Transportation (Mn/DOT). 1991. Noise Analysis: Stop

and Go Traffic Procedures. Noise Group of Environmental Engineering,Engineering Services Section. Minneapolis, MN.

National Economic Research Associates (NERA). 1993. External Costs of Electric

Utility Resource Selection in Nevada, Final Report. Prepared for the NevadaPower Company. Cambridge, MA.

National Highway Traffic Safety Administration (NHTSA). 1996. Traffic Safety Facts

1995. Washington, DC: U.S. Department of Transportation.

National Research Council (NRC). 1991. Rethinking the Ozone Problem in Urban

and Regional Air Pollution. Committee on Tropospheric Ozone Formation andMeasurement. Washington, DC: National Academy Press.

National Safety Council. 1993. Estimating the Costs of Accidents, 1992. 5C893.Itasca, IL: National Safety Council, Statistics Department.

Nelson, Jon P. 1978. Economic Analysis of Transportation Noise Abatement.Cambridge, MA: Ballinger Publishing Company.

Nelson, Paul M. 1987. Transportation Noise Reference Book. London, UK:Butterworth and Company.

Organisation for Economic Co-operation and Development (OECD). 1992. RoadTransport Research Programme. Environmental Impact Assessment of Roads.Paris, FR.

Planco. 1990. Externe Kosten des Verkehrs: Schierre, Strasse, Binnenschiffart.Essen, Germany.

Pope, C. Arden, III, Michael J. Thun, Mohan M. Namboodiri, Douglas W. Dockery,John S. Evans, Frank E. Speizer, and Clark W. Heath, Jr. 1995. “Particulate AirPollution as a Predictor of Mortality in a Prospective Study of U.S. Adults.”Respiratory and Critical Care Medicine, Vol. 151, No. 3 (March), pp. 669–674.

Rothengatter, Werner. 1989. Aspects Economiques. Report to the EuropeanConference of Ministers of Transport Committee of Deputies. Paris, FR.

References 53

Rothengatter, Werner. 1994. “Obstacles to the Use of Economic Instruments inTransport Policy.” In Internalizing the Social Costs of Transport. EuropeanConference of Ministers of Transport (ECMT). Paris, FR: Organisation forEconomic Co-operation and Development, pp. 113–152.

Rudder, F. F., D. F. Lam, and P. Chueng. 1979. User’s Manual: FHWA Level 2

Highway Traffic Noise Prediction Model, STAMINA 1.0. Federal HighwayAdministration, Report No. FHWA–RD–78–138. Washington, DC: U.S.Department of Transportation.

Schwartz, Joel, Douglas W. Dockery, and Lucas M. Neas. 1996. “Is Daily MortalityAssociated Specifically with Fine Particles?” Air and Waste Management

Association, Vol. 46, No. 10 (October), pp. 927–939.

Schwartz, Joel, and Robert Morris. 1995. “Air Pollution and Hospital Admissions forCardiovascular Disease in Detroit, Michigan.” American Journal of

Epidemiology, Vol. 142, No. 1 (July 1), pp. 23–25.

Small, Kenneth A., and Camilla Kazimi. 1995. “On the Costs of Air Pollution fromMotor Vehicles.” Journal of Transport Economics and Policy, Vol. 29, No. 1(January), pp. 7–32.

Spady, Richard H. 1979. Econometric Estimation of Cost Functions for the

Regulated Transportation Industries. New York, NY: Garland Publishing, Inc.

Spady, Richard H., and Ann F. Friedlaender. 1976. Economic Estimation of Cost

Functions in the Transportation Industries. Cambridge, MA: MIT Center forTransportation Studies. Report No. 76–13 (September).

Thurston, George D., Kazuhiko Ito, Carl G. Hayes, David V. Bates, and MortonLippmann. 1994. “Respiratory Hospital Admissions and Summertime Haze AirPollution in Toronto, Ontario: Consideration of the Role of Acid Aerosols.”Environmental Research, Vol. 65, No. 2 (May), pp. 271–290.

Transportation Research Board (TRB). 1989. Providing Access for Large Trucks.Special Report 223. Washington, DC: National Academy Press.

Transportation Research Board (TRB). 1994. Compensating Injured Railroad

Workers Under the Federal Employers’ Liability Act. Special Report 241.Washington, DC: National Academy Press.

Transportation Research Board (TRB). 1995. Expanding Metropolitan Highways:

Implications for Air Quality and Energy Use. Special Report 245. Washington,DC: National Academy Press.


Transportation Research Board (TRB). 1996. Paying Our Way: Estimating Marginal

Social Costs of Freight Transportation. Special Report 246. Washington, DC:National Academy Press.

U.S. Chamber of Commerce. 1991. 1991 Analysis of Workers Compensation

Laws. Washington, DC: U.S. Chamber of Commerce.

U.S. Department of Energy. 1995. Emissions of Greenhouse Gases in the United

States, 1987–1994. DOE/EIA–0573. Energy Information Administration.Washington, DC: Government Printing Office.

U.S. Environmental Protection Agency. 1994a. National Air Pollutant Emission

Trends, 1900–1993. Washington, DC: Government Printing Office.

U.S. Environmental Protection Agency. 1994b. User’s Guide to MOBILE5 (Mobile

Source Emissions Factor Mode 1). Office of Mobile Sources. EPA–AA–AQAB–94–01. Ann Arbor, MI.

U.S. Environmental Protection Agency. 1995. User’s Guide to PART5: A Program

for Calculating Particulate Emissions from Motor Vehicles. Office of MobileSources: EPA–AA–AQAB–94–02. Ann Arbor, MI.

U.S. Environmental Protection Agency. 1997. Final Revisions to the Ozone and

Particulate Matter Air Quality Standards. Office of Air Quality Planning andStandards, U.S. Environmental Protection Agency (Online). Available:http://www.epa.gov/oar/oaqps/ozpmbro/current.htm. April 13, 1998.

U.S. Office of Management and Budget. 1991. Regulatory Program of the United

States Government, April 1, 1991–March 31, 1992. Washington, DC:Executive Office of the President.

Varian, Hal R. 1984. Microeconomic Analysis, 2nd Ed. New York: W. W. Norton& Company.

Verhoef, Erik. 1994. “External Effects and Social Costs of Road Transport.”Transportation Research, Vol. 28A, No. 4 (July), pp. 273–287.

Vickery, William. 1968. “Automobile Accidents, Tort Law, Externalities, andInsurance: An Economist’s Critique.” Law and Contemporary Problems,Vol. 33, No. 3 (Summer), pp. 464–487.

Viscusi, W. Kip. 1993. “The Value of Risks to Life and Health.” Journal of

Economic Literature, Vol. 31, No. 4 (December), pp. 1912–1946.

Wang, Michael Q., and Danilo J. Santini. 1995. “Monetary Values of Air PollutionEmissions in Various U.S. Areas.” Paper presented at the Transportation ResearchBoard 74th Annual Meeting, Washington, DC.

References 55

Weaver, C. S. 1988. Feasibility and Cost-Effectiveness of Controlling Emissions

from Diesel Engines in Rail, Marine, Construction, Farm, and Other Mobile

Off-Highway Equipment. Washington, DC: Radion Corporation.

Willeke, Rainer. 1994. “Germany.” In Benefits of Different Transport Modes.European Conference of Ministers of Transport (ECMT). Paris, FR: Organisationfor Economic Co-operation and Development, pp. 5–37.

Williams, I. D., and I. S. McCrae. 1995. “Road Traffic Nuisance in Residential andCommercial Areas.” Science in the Total Environment, Vol. 169, pp. 75–82.

Documents

External Costs of Truck and Rail Freight Transportation